Hybrid training: what the evidence actually says

Hybrid training means building strength and endurance in the same week, deliberately, without letting either one go backwards. That much is easy to say. The hard part is that almost everything people believe about how to do it traces back to a single study run in 1980 on untrained men — and most of what followed has not held up.

This goes through what the research actually supports, where it is genuinely unsettled, and where the advice everyone repeats turns out to be a coaching habit rather than a finding. Then it puts a week together.

Numbers in brackets point to the evidence section at the end, where every claim here is backed by the study it came from.

First, the fear: does conditioning cost you strength?

This is the question that stops people training both, so it is worth answering carefully rather than reassuringly.

The worry starts with Robert Hickson's 1980 study at the University of Illinois.1 Three groups of untrained men trained for ten weeks: strength only, endurance only, or both. The group doing both kept pace with the strength-only group for seven weeks, then flattened out and went backwards in weeks nine and ten. His conclusion — that training for both at once blunts strength — became the foundation of forty-five years of gym lore.

Now look at what they were actually doing. That group lifted five days a week and did endurance six days a week. Eleven sessions. And the two were separated by as little as fifteen minutes. The endurance work was all-out intervals plus continuous running "as fast as possible." There were seven men in the group.

That is close to the worst-case scenario anyone has ever tested. The late collapse is just as easily explained by ordinary overtraining as by anything special about mixing the two, and the study has no way to tell those apart.

What happens when you look at everything since

The best summary we have pooled forty-three studies and 1,090 people, comparing those who lifted and did conditioning against those who only lifted.2

What was measured	Effect of adding conditioning
Maximal strength	No measurable difference
Muscle size	No measurable difference
Explosive strength — jumping, sprinting, producing force fast	A small but real reduction

So the headline is much narrower than the folklore. How strong you get and how much muscle you build are essentially untouched. The one thing that reliably suffers is explosive power.

If you lift to get strong or to get bigger, conditioning is close to free. If your sport is decided by how fast you produce force — sprinting, jumping, Olympic lifting — you have something real to manage, though it is still a small effect rather than a catastrophe.

One caveat, stated plainly: this research is not great. The studies behind that summary scored an average of 4.3 out of 10 on a standard quality scale. Most ran six to twelve weeks on fewer than thirty recreationally active young men. Data on genuinely elite athletes barely exist.2,3

The one thing that actually mattered

Researchers tested every variable they could think of — how trained people were, their age, whether they ran or cycled, which came first, how many sessions a week. Only one made a consistent difference:

• Both in the same session, back to back: measurable loss of explosive strength
• Separated by three hours or more: no measurable effect at all

A separate analysis found the same thing in trained people — a clear negative effect when the two were done together, essentially nothing when they were a couple of hours apart.3

That is the most useful finding in this entire field, and notice what kind of instruction it is. It is about your calendar, not your program. Put a few hours between your hard lifting and your hard conditioning and most of the measurable problem disappears.

Be honest about the precision, though. The specific numbers you see quoted — three hours, six hours, a full day — come from researchers' judgement calls, not from anyone running a proper comparison. Nobody has tested three hours against six against twenty-four. "Separate them, and more is better" is what the evidence supports. The exact figure is a convention.

The science-y explanation you have heard, and why it is wrong

The popular story goes: cardio switches on an enzyme called AMPK, AMPK blocks a pathway called mTOR, mTOR is what builds muscle, therefore cardio shuts off muscle growth. Clean, plausible, repeated in a thousand articles.

It has mostly failed to survive contact with actual humans.4

Thirty minutes of moderate cycling does not block that pathway. Doing endurance work after lifting does not shut down the muscle-building signal. A study that tracked these proteins across eight weeks of combined training in trained cyclists found no effect at all, at any intensity, and concluded the interference people worry about is not caused by this mechanism. A 2018 review that lays out the whole pathway in detail admits in the same paper that human studies have never actually found the interference it predicts.

The signalling is real in some situations, and much clearer in rats than in people. What keeps failing is the leap from a short-term chemical signal to an actual long-term result. If this switch really governed outcomes, that forty-three-study analysis would not have found muscle growth completely unaffected.

The better explanation is much more boring: you are just tired. You do a hard conditioning session, you are fatigued, your next lifting session is worse than it should have been, and over months you accumulate slightly less good training. That is a recovery problem with a scheduling fix, not a biological incompatibility.

Running versus cycling versus rowing

You will read everywhere that running interferes with lifting and cycling does not. The evidence for that is weaker than its confidence suggests.

A 2012 analysis found running hurt strength and size while cycling did not.5 That is where the claim comes from. But the larger 2022 analysis found the opposite — cycling was the one that reduced explosive strength, running was not — and even that reversed when one unusual study was removed from the pool.2

Two analyses, opposite answers, one of them fragile to a single study. The honest position is that nobody has shown this convincingly in either direction.

There is still a good reason to lean on the bike or the rower when you are lifting heavy, and it has nothing to do with any of that: running beats your legs up more. More impact, more muscle damage, more soreness for the same amount of cardio. Wrecked legs make for a bad squat session. That is a fatigue argument, and it fits everything else here. It just isn't the argument people think they're making.

The direction nobody worries about: lifting makes you better at endurance

The whole conversation runs one way, which is odd, because the effect running the other way is better evidenced and considerably larger.

Pooling controlled trials in highly trained middle- and long-distance runners, adding strength training produced a large improvement in running economy — how much energy it costs you to hold a given pace.6 Eight to twelve weeks, lifting two or three times a week, in people who were already competitive runners.

A much bigger review across thirty-one studies and 652 runners confirmed it and added detail: heavy lifting worked, light lifting did not. Submaximal loads and isometric work showed nothing. A companion analysis of thirty-eight studies found heavy strength training improved actual race performance, not just efficiency.7 The same pattern shows up in cyclists.8

Here is the detail that matters most, and it holds across all of it: lifting improves endurance performance without improving your VO₂max. Every one of those analyses found no change in maximum oxygen uptake. You are not building a bigger engine. You are reducing the drag — stiffer tendons, better force application, less energy wasted with every stride.

So the trade is lopsided, and it favours doing both. Conditioning costs you a little explosive power. Lifting buys you a large improvement in how efficiently you move. And your aerobic fitness is not blunted by lifting at all — VO₂max improvements in people doing both match people doing endurance alone almost exactly.2

Structuring the conditioning

The 80/20 rule, and what survives of it

You have heard that elite endurance athletes do about 80% of their training easy and 20% hard. That part is true and well documented across skiers, cyclists, rowers, swimmers and marathoners — the easy share runs somewhere between 70% and 88%.9

What does not survive is the fancier claim built on top of it. Coaches argue about whether you should go "polarised" (lots of easy, some very hard, almost nothing in the middle) or "pyramidal" (lots of easy, a decent chunk of moderate, less very hard).

Look at the elite data again and most of it is pyramidal. Professional cyclists, elite rowers, elite swimmers — in all of them the moderate zone is bigger than the hard zone, which is the definition of pyramidal, not polarised.9 And an analysis of 151,813 marathons by 119,452 runners found pyramidal was the most common pattern among the fastest runners.10

The controlled research lands in the same place. The highest-quality comparison anyone has run — pooling the individual data from thirteen studies rather than just their summaries — found no difference between the two approaches on either VO₂max or race performance.11 A separate analysis of forty-one studies in trained cyclists found nothing either.12

There is also a measurement problem that quietly undermines the whole debate. One study tracked twenty-nine elite skiers across 570 sessions and scored the same training two different ways. Counted by heart-rate zones it looked like 96% easy. Counted by what the session was actually meant to be, it looked like 87% easy. Heart rate lags, so it systematically under-counts hard work — by about 28%.13 Whether your training is "polarised" is partly an accident of how you measure it.

Zone 2 is useful. It is not magic.

Zone 2 has been sold for several years as a uniquely powerful stimulus for mitochondria and fat burning. A 2025 review set out to test that claim specifically and found no substantive evidence that Zone 2 is better than higher intensities for either one.14

Several of the specific mechanisms do not hold up. More than three hours of Zone 2 work does not change the energy-sensing molecules in muscle that are supposed to trigger the adaptation. It does not raise AMPK in trained people. The calcium pathway that gets cited constantly has, as far as that review could find, never actually been measured during Zone 2 exercise at all.

There is a definitional mess too. Zone 2's upper limit ranges from roughly a quarter to four-fifths of your maximum depending on your fitness. In an unfit person it is under 100 watts on a bike; in a trained endurance athlete it is around 300. It is not one intensity, it is a relationship. And the heart-rate charts everyone uses — "Zone 2 is 60 to 70% of max" — have no real physiological basis and will land different people in completely different states.14

So what is easy work actually for? It is cheap. After an hour of easy work, a trained athlete's nervous system is recovered in five to ten minutes. After a hard session it takes around thirty.15 Easy work lets you pile up training without piling up fatigue — which is precisely what you need when you are also trying to lift. There is also decent evidence that easy endurance work is the best stimulus for growing new capillaries specifically, better than intervals, though that finding is contested.16

That is a good enough reason to do plenty of it. It is not the same as it being magic.

How much hard work, and what kind

Here the research gets unusually practical. A 2026 analysis of sixty-nine randomised trials mapped out exactly how much you get for the time you put in.17

To get this much of the available benefit	Intervals need	Steady moderate cardio needs
Half of it	About 6 minutes of work	36 minutes
Most of it (80%)	About 11 minutes of work	About 52 minutes
All of it	About 23 minutes of work	63 minutes

Two things fall out of that. Intervals get you most of the way in about a fifth of the time. And the interval curve flattens out — past roughly twenty-three minutes of actual hard work in a session, you are paying fatigue for almost nothing. Both methods end up in the same place. Intervals just get there faster.

For how to build the intervals: longer beats shorter. Comparing hard intervals against all-out sprint intervals, there was no difference overall — but intervals of four minutes or longer beat short sprints for actual race performance.18 Short sprints are the most time-efficient way to build mitochondria per hour, but they produced no capillary growth and no VO₂max advantage.16,18

So if you get one hard conditioning session a week, make it four to six minute efforts with two to four minutes rest between. Not thirty-second sprints.

The "middle intensity trap," handled honestly

The standard line is that moderate-hard work is a dead zone — too hard to count as recovery, too easy to drive real adaptation. Treat this as a useful rule of thumb rather than an established fact, because the direct evidence is one study with six people per group.19

That study held the hard work and total volume constant and varied only how the rest was split. The easy-emphasis group improved their 10 km time by about two and a half minutes against two minutes for the moderate group. Suggestive. Also twelve runners.

Against it: the most successful distance running method currently in use — the Norwegian approach behind Jakob Ingebrigtsen's Olympic title and indoor world record — is built on three to four moderate-intensity sessions a week, on purpose, with blood lactate measured during the session to keep the intensity controlled.20 That is a lot of time in the supposed dead zone. Meanwhile pyramidal training, which by definition contains more moderate work, shows no penalty in any of the pooled comparisons.11,12

The defensible version is narrower, and still worth following: if most of your training ends up at a vague moderately-hard effort, you are wasting a limited budget. You pay real recovery for it and then cannot do your genuinely hard sessions properly. That is an argument about where your hard work goes, not about the moderate zone being useless.

Structuring the lifting

Load: heavy for strength, flexible for size

Getting stronger and getting bigger have genuinely different requirements, and confusing them is where most hybrid programs go wrong.

Pooling twenty-eight trials: for muscle size, how heavy you lift barely mattered. For maximal strength, it mattered a great deal — heavy loading came out ahead with about 98% certainty.21 A separate analysis of twenty-one studies found the same split: considerably bigger one-rep-max gains on heavy loads, no meaningful difference in muscle growth.22 The largest review of all — nearly 300 studies across both outcomes — ranked heavy loads, multiple sets, three times a week at the top for strength, while noting that almost all the comparisons it made overlapped enough that the rankings should be read loosely.39

The useful range for building muscle is wide but not unlimited. A study that had people train both arms at different loads found roughly equal growth at 40%, 60% and 80% of maximum — and markedly less at 20%.23 So call it 40 to 80%, with sets taken close to failure.

One honest note on why heavy works for strength. The usual explanation involves motor unit recruitment and nervous system adaptation. That is a plausible story with thin direct evidence. What the data actually show is more interesting: heavy lifting improves your one-rep max but does not improve isometric strength any more than light lifting does.22 That gap suggests a lot of it is simply skill — you get good at the specific thing you practise — rather than some deeper physiological change.

Volume: strength plateaus early, size keeps going

A 2025 analysis of sixty-seven studies gave the clearest picture yet, and the two goals behave completely differently.24

Muscle growth keeps responding to more sets, with steadily diminishing returns and no ceiling anyone has found. But the cost climbs steeply. Early on, roughly six extra sets per week buys you another noticeable increment. Once you are already doing thirty sets, it takes about twelve and a half.

Strength is different. It plateaus. One set a week is enough to make progress, two sets is highly efficient, and past about five sets per lift per week the returns become unreliable. What does keep paying is frequency — adding another session per week helped substantially, with near-total statistical confidence. For muscle growth, frequency barely registered.

For a hybrid athlete that is close to a gift. Strength is the cheaper adaptation. A handful of heavy sets spread across more days gets you most of what is available, leaving budget for conditioning.

Training to failure: the recovery cost decides it

For muscle growth there is a shallow but real benefit to pushing sets closer to failure — roughly half a percent more growth per rep closer. For strength there is essentially none; across fifty-five studies, how close you trained to failure made no difference to strength gains at all.25

The cost side is where this gets decided. One study compared sets taken to failure against the same total volume stopped short:26

• Stopped short of failure: fully recovered within 6 hours
• Taken to failure: still impaired at 24 and 48 hours, fully recovered only at 72 hours

Identical volume. Twelve times the recovery cost.

And it is not just about recovery. In an eight-week squat study, the group that stopped their sets earlier did 40% fewer total reps and gained more strength — and nearly three times the jump height improvement.27

If lifting is all you do, training to failure is a reasonable trade for the extra growth. If you also need to run hard on Thursday, it is close to the worst purchase on the menu: a fraction of a percent of muscle for three days of degraded performance in everything else.

Do you need to test your max?

Mostly no, and the alternatives are better than their reputation.

Judging how many reps you had left — "reps in reserve" — is a valid way to regulate your training, with a known error bar. Across twelve studies and 414 people, lifters underestimated how many reps they had left by about one rep on average. Accuracy gets better the closer you are to failure and gets noticeably worse past about twelve reps. Interestingly, training experience made almost no difference to accuracy.28

So judging effort is most reliable exactly where strength work lives: heavy loads, low reps, close to failure. It is least reliable in the high-rep moderate-load range.

The alternatives carry more error than people assume. Estimating your max from bar speed is off by around 10% — roughly 40 lb on a squat — and systematically guesses high.29 Percentage charts are badly calibrated at moderate loads: at 60% of your max, predictions of how many reps you have left were off by about three, improving to well under one rep at 80%.30

Practical upshot: judge your effort, anchor it on heavy sets of the big lifts where that judgement is most reliable, and distrust percentage charts in exactly the rep ranges where people lean on them hardest. Test a true max when the number itself matters — a competition, a real benchmark — not on a schedule.

Where metcons actually fit

Circuit-style conditioning — the CrossFit-descended workout where you move weight fast against the clock — sits awkwardly in hybrid training, mostly because people use it as a substitute for things it doesn't replace.

Start with what it actually taxes. Researchers measured the energy systems during a two-minute benchmark workout in fourteen highly trained athletes: 45% from the glycolytic system, 40% aerobic, 15% from stored phosphates.31 Across a broader range of these workouts, heart rate runs very high and blood lactate climbs to several times resting levels — but oxygen uptake sits at only 57 to 66% of maximum.32

Read that combination carefully. It feels brutal, lactate is enormous, and yet the aerobic demand is moderate. These are lactate-tolerance sessions. They are too intense to build aerobic volume and too fatiguing to be quality strength work.

Which makes the substitution question answerable:

• They don't replace heavy lifting. Getting maximally strong needs heavy loads and practice at the specific lift. Metcon barbell work is light, high-rep and done exhausted. Tellingly, in competitive CrossFit athletes, overall strength strongly predicted how well they did on these workouts — while VO₂max and anaerobic power predicted nothing at all.33
• They don't replace your aerobic base. Moderate oxygen demand with sky-high lactate is the opposite of what builds aerobic fitness.
• They do train something real — the ability to keep producing power when you are already wrecked. That is exactly what hybrid races test, and it's what easy running and heavy triples both miss.

Be sceptical of strong claims in either direction, because the research is poor. A 2018 pooled analysis across forty-three different measures found no significant effects on anything, including VO₂max, with 94% of the included studies at high or moderate risk of bias.32 A 2025 analysis reported large benefits for strength, power and endurance — but mostly compared against people doing nothing at all, with every finding rated low confidence.34 "Better than sitting on the couch" is not the same as "as good as proper training."

On injuries, the panic is overstated. Pooled rates come out around 3.2 injuries per 1,000 hours of training, with shoulders, spine and knees most affected. Weightlifting runs 2.4 to 3.3 and powerlifting 1.0 to 4.4 over the same hours.35 It's a strength sport with a strength sport's injury profile. One finding is worth acting on, though: in a survey of 386 participants, the injury rate was 14.6% when a coach was present all the time against 31% when a coach was only sometimes there.36

Sensible use: one of these a week, treated as your hard conditioning session. Not as your strength work, and not as your aerobic base.

What a good structure looks like

Everything above collapses into a short list of rules, in rough order of how much they matter.

1. Put a few hours between hard lifting and hard conditioning. This is the only variable that showed up consistently across every analysis. If you can only train once a day, use different days instead.
2. One or two genuinely hard conditioning sessions a week. Not four. Eleven minutes of interval work gets you most of what's available; twenty-three gets you all of it. Past that you're buying fatigue.
3. Most of your conditioning should be easy enough to hold a conversation. Not because it's magic, but because it's the only training load that doesn't compete with your lifting for recovery.
4. Keep heavy work heavy and keep the volume low. Strength plateaus around five sets per lift per week and responds better to more days than more sets.
5. Stop most sets one to three reps short of failure. Six hours of recovery instead of seventy-two, for a fraction of a percent of muscle. In a hybrid week this isn't close.
6. Put the priority first — first in the week, first in the session. Lifting before conditioning gives a small edge for strength, but this matters far less than separating them.
7. Something has to lead. You can hold two qualities and push one. Pushing both at once usually pushes neither.

A worked week, four to five days

Day	Session	Why it's there
Monday	Lower body, heavy. 2–4 sets per main lift, stopping 2–3 reps short.	Fresh from the weekend. Priority work on the freshest day.
Tuesday	Easy aerobic, 40–60 min. Conversational.	Almost free in recovery terms. Builds the base without competing.
Wednesday	Upper body.	Cheap to recover from. Sits comfortably near conditioning.
Thursday	Hard intervals. 4–6 × 4 min, 2–4 min rest. About 20 min of work.	Two days clear of the heavy leg day. Long intervals beat short ones.
Friday	Rest, or easy aerobic.	The adaptation happens here.
Saturday	Second lower body day, or a metcon.	Whichever quality is leading this block.
Sunday	Rest.	Not optional.

Notice what that week does not have. No day where a heavy squat session and hard intervals sit back to back. No third hard conditioning session. No sets to failure the day before hard running. The structure is doing the work, not the heroics.

When one side has to give

Sometimes a race block or a strength block means something gets squeezed. The research here is genuinely reassuring.

One study took trained adults through sixteen weeks of lifting, then cut them to a maintenance dose for thirty-two weeks. One session a week — even a single set — held onto their strength, and actually added another 7 to 8% on top.37 Younger adults also kept their muscle size. Older adults did not, which is worth knowing if you're past sixty.

Broader reviews land in the same place: strength and muscle can be held for up to eight months on as little as one session a week and one set per exercise, and endurance for up to fifteen weeks on two sessions a week.38 The non-negotiable variable in every case is how heavy you lift. Cut frequency, cut volume. Don't cut the load.

What to be sceptical of, including some of this

A fair reading of this research means admitting how thin parts of it are.

Most studies on training both run six to twelve weeks on fewer than thirty recreationally active young men, at an average quality score of 4.3 out of 10. The training-distribution research scores 4.5 out of 7 with acknowledged publication bias. The entire CrossFit-style literature is rated low confidence and compares mostly against people doing nothing. Data on genuinely elite athletes barely exists anywhere in this.

Several findings are contested between good studies rather than settled. Whether running or cycling interferes more: two analyses, opposite answers. Whether being more trained makes interference worse: one found yes, a larger one found no. Whether easy work builds capillaries: one large analysis says it's the best stimulus, another says it does nothing. Whether metcons improve VO₂max: individual studies report real gains, the pooled analysis finds nothing.

And some of what gets presented as science is just convention. "Test your max every eight to twelve weeks." "You need twenty-plus sets per muscle per week." "Two hard sessions a week is the ceiling." None of those has direct experimental backing at the level of the research around them. They might still be good advice. They are not findings.

What does hold up well enough to build on: separate your hard sessions, keep most conditioning easy and some of it genuinely hard, keep heavy work heavy and low in volume, stay short of failure, and let one quality lead at a time. That is a short list, and it is most of the game.

How SOFReady handles it

The Hybrid track is built around exactly the structure above — strength and conditioning in the same week, sequenced so the hard sessions don't collide, progressed across blocks that build on each other instead of resetting every month.

Prescriptions run on perceived effort rather than assuming you know your current maxes, with percentages reserved for the main barbell lifts where they're most reliable. Programming adapts to the equipment you actually have. And when one quality needs to lead for a while, the Strength and Engine tracks exist for that, with the chain designed so you can move between them without starting over.

If you want the longer treatment of either side, strength and conditioning goes deeper on the lifting, and hybrid race prep covers building toward a dated event.

The evidence

Every claim above, with the study behind it and the numbers as reported. Effect sizes are standardised mean differences (SMD) unless noted; a value near zero means no meaningful difference, and the p-value is the probability the result is noise — below 0.05 is the usual bar for calling something real.

1. Hickson RC (1980). European Journal of Applied Physiology 45:255–263. Ten weeks, untrained men, roughly 7–8 per group. Concurrent group: strength 5 d/wk, endurance 6 d/wk, modes separated by 15–120 min. Leg strength matched the strength-only group for 7 weeks, then levelled off and declined in weeks 9–10. VO₂max rose ~20–25% in both endurance groups.
2. Schumann M et al. (2022). Sports Medicine 52(3):601–612. 43 studies, 1,090 participants. Maximal strength SMD −0.06 (95% CI −0.20 to 0.09, p=0.446); hypertrophy SMD −0.01 (p=0.919); explosive strength SMD −0.28 (−0.48 to −0.08, p=0.007). Only significant moderator was session separation: same session SMD −0.31 (p=0.043); ≥3 h apart, not significant. Running vs cycling, training status, age and session order all non-significant. Mean study quality 4.3/10 on PEDro.
3. Petré H, Hemmingsson E, Rosdahl H, Psilander N (2021). Sports Medicine 51:991–1010. 27 studies, 750 participants. In trained individuals: same session (<20 min apart) ES −0.66 (p<0.01); separated >2 h, ES −0.10 (p=0.55). Also reported a training-status gradient — untrained 0.03, moderately trained −0.20, trained −0.35 — which Schumann (2022) did not replicate.
4. On the AMPK–mTOR hypothesis. Jones TW et al. (2022), Frontiers in Physiology 13:852595 — 8 weeks concurrent training in trained cyclists, no effect on phosphorylation of AMPK or mTOR pathway proteins at any endurance intensity; concluded interference is "not mediated by the mTOR-AMPK axis." Murach KA & Bagley JR (2016), Sports Medicine 46(8) — resistance-induced mTORC1 signalling not impaired by subsequent endurance exercise. Methenitis S (2018), Sports 6(4):127 — presents the pathway, then concedes human studies "have not observed such molecular 'interference'." Baar K (2014), Sports Science Exchange 27(136) — 30 min moderate cycling did not inhibit mTOR; AMPK could not explain the inhibition that was seen.
5. Wilson JM et al. (2012). J Strength Cond Res 26(8):2293–2307. 21 studies, 422 effect sizes. Running-concurrent reduced hypertrophy (0.68 vs 1.54 strength-only) and strength (1.23 vs 2.22), p<0.05; cycling did not. Note: compares within-group pre-post effect sizes across arms rather than matched between-group comparisons, and the modality finding is contradicted by Schumann (2022).
6. Balsalobre-Fernández C, Santos-Concejero J, Grivas GV (2016). J Strength Cond Res 30(8):2361–2368. 5 controlled trials, 93 competitive middle/long-distance runners. Running economy SMD −1.42 (95% CI −2.23 to −0.60). Protocols: 8–12 weeks, 2–3×/wk, 40–70% 1RM.
7. Llanos-Lagos C et al. (2024). Sports Medicine 54:895–932 and 54(7):1801–1833. Economy: 31 studies, 652 runners — high load (≥80% 1RM) ES −0.266 (p=0.039); combined methods ES −0.426 (p=0.018); submaximal and isometric not significant. Performance: 38 studies, 894 participants — high load ES −0.469 (p=0.029); combined ES −1.035 (p=0.036). No improvement in VO₂max in either (p>0.072). Certainty rated very low to moderate.
8. Llanos-Lagos C et al. (2025). Eur J Appl Physiol 126(1). 17 studies, 262 cyclists. Cycling efficiency ES 0.353 (p=0.012); time-trial performance ES 0.463 (p=0.016); no effect on VO₂max. Authors state low certainty of evidence.
9. Stöggl T & Sperlich B (2015). Frontiers in Physiology 6:295. Review of observed elite distributions (easy / moderate / hard): pro cyclists 88/11/2; elite rowers 70–77/15–22/6; elite swimmers 77/12/11; elite XC skiers 86/5.3/3.3; top marathoners 78/4/18. Most are pyramidal (moderate > hard), not polarised.
10. Muniz-Pumares D et al. (2024). Sports Medicine 54(4):1023–1035. 151,813 marathons by 119,452 runners, 16 weeks pre-race. Pyramidal was the most common distribution in >80% of the fastest runners. Marathon time correlated negatively with total volume and easy-zone proportion (R²≥0.90). Observational; zones derived from consumer wearables.
11. Rosenblat MA et al. (2025). Sports Medicine 55(3):655–673. Network meta-analysis of individual participant data — 13 studies, 348 participants. Polarised vs pyramidal: VO₂max SMD −0.06 (p=0.68); time trial SMD −0.05 (p=0.34). A level interaction (SMD −0.63, p<0.05) suggested competitive athletes may do better polarised and recreational athletes pyramidal.
12. Cove B et al. (2025). J Sci Med Sport 28(5):423–434. 41 studies, 81 training groups, 797 trained cyclists. VO₂max: polarised g=0.52 vs non-polarised g=0.38, not significant. Time trial: 0.31 vs 0.40, not significant. Intervention duration predicted gains (p≤0.008); weekly and total volume predicted nothing (p>0.16).
13. Sylta Ø, Tønnessen E, Seiler S (2014). Int J Sports Physiol Perform 9(1):100–107. 29 elite XC skiers, 570 sessions. Same training scored 96.1/2.9/1.1 by heart-rate time-in-zone vs 86.6/11.1/2.4 by session goal. Time-in-zone underestimated high-intensity training by 27.5%.
14. Storoschuk KL et al. (2025). Sports Medicine. Narrative review targeting the Zone 2 claim directly. "Failed to uncover substantive evidence supporting claims that Zone 2 is superior to higher exercise intensities for improving mitochondrial and fat oxidative capacity." 200 min of Zone 2 does not change muscle ATP/ADP/AMP; does not raise AMPK in endurance-trained men; no studies have measured intracellular calcium during Zone 2 exercise. Zone 2's upper bound ranges from ~24% to ~80% of VO₂max depending on training status.
15. Seiler S, Haugen O, Kuffel EE (2007). Med Sci Sports Exerc 39(8):1366–1373. 17 trained and highly trained runners. Heart-rate variability returned to baseline 5–10 min after 60 or 120 min below the first threshold; ~30 min after threshold or above-threshold work in highly trained athletes; ~90 min after hard work in the less-trained group.
16. Mølmen KS, Almquist NW, Skattebo Ø (2024). Sports Medicine. 353 studies, 5,973 participants. Mitochondrial content: endurance 23±5%, intervals 27±5%, sprints 27±7% — no significant difference. Capillary density: endurance 13±3%, intervals 7±4%, sprints no significant increase (endurance significantly greater, p<0.05). Frequency mattered: 6 > 4 > 2 sessions/week. Note: one other meta-analysis cited by Storoschuk reports no capillary effect from low-intensity work — this is contested.
17. Mattioni Maturana F et al. (2026). Sports Medicine. 69 randomised trials, 2,387 participants. HIIT superior to moderate continuous training by 1.20 ml/kg/min (95% CI 0.83–1.57). HIIT dose-response plateaus: 50% of maximal effect at 5.8 min of work per session, 80% at ~11 min, 100% at 23.2 min. Moderate continuous is linear: 50% at 36 min, 80% at ~52 min, 100% at 63 min. Ceilings nearly identical (4.31 vs 4.13 ml/kg/min).
18. Rosenblat MA, Perrotta AS, Thomas SG (2020). Sports Medicine 50:1539–1552. 6 trials, moderately-to-highly trained men. HIIT vs sprint intervals on time trial: +0.9% (not significant). Long intervals (≥4 min) vs sprint intervals: +2.0% (90% CI 0.7–3.3, p=0.01, ES 0.88). No difference in VO₂max between any group. Recommended 4–6 min work bouts with 2–4 min recovery.
19. Esteve-Lanao J, Foster C, Seiler S, Lucía A (2007). J Strength Cond Res 21(3):943–949. 12 subelite runners, 6 per group, 5 months. Volume and hard-zone work matched; easy-emphasis group 80.5/11.8/8.3 vs moderate-emphasis 66.8/24.7/8.5. 10.4 km improvement: −157±13 s vs −121.5±7.1 s (p=0.03). Six per group.
20. Casado A, Foster C, Bakken M, Tjelta LI (2023). Int J Environ Res Public Health 20(5):3782. Description of the Norwegian double-threshold method: 3–4 lactate-guided threshold sessions plus one VO₂max session weekly within 150–180 km/week, lactate held to 2.0–4.5 mmol/L. Distribution pyramidal in preparation. Case studies, not a controlled trial.
21. Lopez P et al. (2021). Med Sci Sports Exerc 53(6):1206–1216. 28 RCTs, 747 participants. Hypertrophy, high vs low load: SMD 0.12 (−0.06 to 0.29, p=0.241). Strength, high vs low: SMD 0.60 (0.38–0.82, p<0.001), with heavy load having 98.2% probability of being the best option.
22. Schoenfeld BJ, Grgic J, Ogborn D, Krieger JW (2017). J Strength Cond Res 31(12):3508–3523. 21 studies, all sets to failure. 1RM: heavy ES 1.69 (+35.3%) vs light ES 1.32 (+28.0%), p=0.003. Isometric strength: 0.64 vs 0.55, p=0.43. Hypertrophy: 0.53 vs 0.42, p=0.10 (p=0.56 at study level).
23. Lasevicius T et al. (2018). Eur J Sport Sci 18(6):772–780. Within-subject, volume-equated, 12 weeks, sets to failure. Quadriceps cross-sectional area: 20% 1RM +8.9%; 40% +20.5%; 60% +20.4%; 80% +19.5%. Elbow flexors: 11.4 / 25.3 / 25.1 / 25.0%.
24. Pelland JC et al. (2025). Sports Medicine 56(2). Bayesian analysis, 67 studies, 2,058 participants. Hypertrophy: +0.24% per added weekly set (95% credible interval 0.15–0.33), square-root curve, no identified plateau; roughly 6 extra sets per further increment at low volumes rising to ~12.5 at high volumes. Strength: +0.21% per set but reciprocal curve with a functional plateau; past ~5 weekly sets per lift, gains inconsistent. Frequency: strength +3.27% per added session (100% posterior probability); hypertrophy +0.32% (91.3% probability, treated as unresolved).
25. Robinson ZP et al. (2024). Sports Medicine 54(9):2209–2231. 55 studies, 140 hypertrophy and 243 strength effects. Hypertrophy declined ~0.48% per additional rep left in reserve (β=−0.019, 95% CI −0.035 to −0.004). Strength: β=0.003 (−0.012 to 0.018) — no relationship. Authors describe overall model fit as modest.
26. Morán-Navarro R et al. (2017). Eur J Appl Physiol 117:2387–2399. 75% 1RM. Jump performance impaired at 0, 6, 24 and 48 h after sets to failure, recovering only by 72 h; volume-matched non-failure protocols recovered by 6 h. Creatine kinase elevated at 6 and 24 h after failure only.
27. Pareja-Blanco F et al. (2017). Scand J Med Sci Sports 27(7):724–735. 8 weeks of squat training, 20% vs 40% velocity loss per set. The 20% group did 185.9±22.2 reps vs 310.5±42.0 (40% fewer) yet gained more 1RM (+18.0% vs +13.4%) and far more jump height (+9.5% vs +3.5%).
28. Halperin I et al. (2022). Sports Medicine 52:377–390. 12 studies, 414 participants, 262 effect sizes. Participants under-predicted reps to failure by 0.95 reps (95% CI 0.17–1.73). Accuracy improved 0.025 reps per percentage point closer to failure and degraded beyond 12 reps (−0.47 reps per extra rep). Training experience had no meaningful effect.
29. Greig L et al. (2023). Sports Medicine 53(9):1693–1708. 26 studies, 641 participants, 137 velocity-based 1RM prediction models. Aggregate standard error 9.8% (95% CI 7.4–12.2); back squat 18.6 kg / 12.3%. Systematic overestimation of +4.5 kg (+3.7%). No accuracy gain from using more than two loads or individualised velocity thresholds.
30. Qin X, Liu B, García-Ramos A (2025). BMC Sports Sci Med Rehabil 17:60. Bench press. Error in predicted reps-in-reserve using percentage-of-1RM tables: 2.9±3.5 reps at 60% 1RM, improving to 0.6±1.2 reps at 80%. A velocity-based method was more accurate (0.4±1.2 reps at 60%).
31. Rios M et al. (2024). Sensors 24(2):513. "Isabel" (30 snatches at 61 kg), 14 highly trained male CrossFit athletes, 117±10 s duration. Energy contribution: oxidative 40%, glycolytic 45%, phosphagen 15%. Peak VO₂ 47.2±4.7 ml/kg/min.
32. Claudino JG et al. (2018). Sports Medicine – Open 4:11. 31 studies reviewed, pooled n=3,597. Across 43 variables including VO₂max and body composition, "no significant results were found for any of the variables." Measured workout intensity: heart rate 54–98% of max, blood lactate 6–15 mmol/L, but VO₂ only 57–66% of max. Mean CONSORT score 37%; 94% of studies at high or moderate risk of bias.
33. Butcher SJ et al. (2015). Open Access J Sports Med 6:241–247. 14 competitive CrossFit athletes. Whole-body strength correlated with benchmark performance (r=−0.88 on one, r=−0.65 on another); VO₂max and anaerobic power predicted nothing.
34. Wang X et al. (2025). BMC Public Health 25:528. 19 studies, 911 participants aged 10–23. Versus mostly inactive controls: strength ES 1.380, power 1.320, endurance 1.820. All outcomes GRADE-rated low certainty; comparisons against alternative training modes were not the primary analysis.
35. Hülsmann M et al. (2021). Dtsch Z Sportmed 72:351–358. 32 studies qualitative, 14 in meta-analysis, 9,306 participants. Pooled injury incidence 3.20 per 1,000 training hours (95% CI 2.06–4.34); prevalence 30.3%. Spine 26.8%, shoulder 25.9%, knee 15.8%. Comparable to weightlifting (2.4–3.3) and powerlifting (1.0–4.4) per 1,000 h.
36. Weisenthal BM et al. (2014). Orthop J Sports Med 2(4). Survey of 386 participants, 6-month recall. 19.4% injured. Trainer present all the time: 14.6% injury rate; present some of the time: 31% (p=0.028). Self-reported, subject to recall bias.
37. Bickel CS, Cross JM, Bamman MM (2011). Med Sci Sports Exerc 43(7):1177–1187. 70 adults. 16 weeks of training (3 d/wk), then 32 weeks at maintenance dose. Both maintenance doses — one session a week with 3 sets, and one session a week with 1 set — preserved 1RM and added a further 7–8% in both age groups. Younger adults maintained muscle fibre size on both doses; older adults did not.
38. Spiering BA, Mujika I, Sharp MA, Foulis SA (2021). J Strength Cond Res 35(5):1449–1458. Review. Endurance performance maintained up to 15 weeks at 2 sessions/week; strength and muscle size maintained up to 32 weeks at 1 session/week and 1 set per exercise in younger populations; older populations require roughly 2 sessions/week and 2–3 sets. Intensity is the variable that must be preserved.
39. Currier BS et al. (2023). Br J Sports Med 57(18):1211–1220. Bayesian network meta-analysis. Strength: 178 studies, 5,097 participants. Hypertrophy: 119 studies, 3,364 participants. Highest-ranked for strength was heavy load, multiple sets, 3×/week (SMD 1.60). Authors note 91% of between-prescription comparisons had overlapping credible intervals, so rankings should be read loosely.