The highest-value fix on this list isn't pressing more often — it's ending every work set one or two reps short of failure. It's the one tip here that doesn't turn out to be a proxy for total volume: in Robinson's 2024 meta-regressions, strength gains held up across a wide range of stopping points. The compressed version: press two to three times a week, stop every work set at 1-2 reps in reserve, keep your setup and bar path identical rep to rep, and add 5 lb only when every work set hits the top of its rep range.
I made a reel with eight overhead press tips and about 51,000 people watched it. This is the long version with the studies attached — and it does not go well for the reel. Three of the eight get walked back outright, including the one I called "this is huge," and only one survives untouched. I went looking for the trial behind the push press tip. There isn't one.
Why is my overhead press not going up?
Because you finish every set with a grinding rep, and for strength specifically that rep buys you nothing.
First, check whether the number is even a problem: whether 135 is a good overhead press and where it ranks at your bodyweight.
Robinson 2024 in Sports Medicine ran multilevel meta-regressions treating proximity to failure as a continuous variable, measured in reps in reserve. In their words: in all of the best-fit models for strength, the confidence intervals of the marginal slopes for estimated RIR contained a null point estimate. The authors call the analysis exploratory and rate the overall model fit as modest — read it as a strong hint, not a settled number.
Grgic's failure meta-analysis in the Journal of Sport and Health Science pooled 15 studies, all in young adults, and found no significant difference between failure and non-failure training for strength: ES = −0.09 (95% CI −0.22 to 0.05). In the studies that didn't equate volume, the effect significantly favoured stopping short — ES = −0.32 (95% CI −0.57 to −0.07).
The honest limit is real: both papers say the opposite for size. Robinson's hypertrophy models had negative marginal slopes whose confidence intervals excluded the null — growth increased as sets ended closer to failure. Grgic's resistance-trained subgroup favoured failure for hypertrophy, ES = 0.15 (95% CI 0.03-0.26). So 1-2 RIR is right when the number on the bar is the goal; if what you want is bigger delts, the calculus shifts. Full case: should you train to failure.
How often should you overhead press?
Two to three times a week — but frequency isn't the lever, it's the delivery system for volume.
Grgic's 2018 meta-analysis in Sports Medicine pooled 22 studies and found strength effect sizes rising with frequency: 0.74 at once a week, 0.82 at twice, 0.93 at three times, 1.08 at four or more (p = 0.003). Two subgroups point straight at this lift. Upper body was significant (p = 0.004) while lower body was not (p = 0.070); multi-joint 1RM tests were significant (p < 0.001) while single-joint tests were not (p = 0.324). The overhead press is upper-body and multi-joint — both subgroups where the effect held.
The counterweight comes from the same paper: with volume equated, there was no significant effect of frequency (p = 0.421), and most pooled studies used untrained participants. Schoenfeld 2019 pooled 25 studies on frequency and hypertrophy and found no significant volume-equated difference either.
So here's the first one I got wrong. My reel said frequency is everything. It isn't. More days is how you fit more quality sets into a week without one session turning into a 90-minute grind — the same mechanism that applies to every muscle group. I made this argument for the squat too, and it's stronger here: there, the lower-body subgroup never reached significance.
What's the right overhead press setup — bar position, elbows, and grip?
Bar racked on your upper chest, elbows just in front of the bar rather than flared wide, and a grip width you don't change from week to week. Grip width is the part of that with measured consequences.
Evangelista 2025 in the Journal of Functional Morphology and Kinesiology built a link-chain model of the trunk and upper limbs and simulated the press. The sticking region appeared when shoulder torque dropped below a critical threshold, and in the static analysis grip width and barbell constraint shaped that sticking region more than muscle architecture parameters did — where your hands sit moves the sticking point more than pennation angle or fascicle length does. The model held up against real barbell kinematics: waveform similarity 0.962 (p = 0.028) before the sticking point, 0.949 (p = 0.014) through it, and no significant match after it. Two caveats — it's a simulation, and the validation data came from maximal bench press trials. That press has its own measured grip line — 1.5 times biacromial width — in how to increase your bench press.
Position and implement are worth real weight too. Saeterbakken and Fimland 2013 in the Journal of Strength and Conditioning Research tested 15 healthy men across four shoulder press variations. Sitting down was worth about 10% more load than standing with the same dumbbells (p < 0.001), and dumbbells cost about 7% against a barbell standing (p = 0.002). Change either one and this week's number isn't comparable to last week's.
The elbow cue I can't back with a study. No trial tests "elbows slightly in front of the bar" against anything — it's a cue with a plausible mechanism, not a tested intervention.
Pressing behind your head instead? Here's who the behind the neck press is safe for.
Does the push press actually increase your strict overhead press?
Nobody knows, because nobody has run the trial. This is the second one I got wrong, and it's the one I oversold hardest.
Soriano's 2019 review in Sports Medicine is the only review of weightlifting overhead pressing derivatives — push press, push jerk, split jerk. The search turned up only 13 articles, nearly all kinetics and kinematics, and the authors close by naming the gap: there is a lack of information regarding the longitudinal training effects of using these lifts. The literature describes what the push press is, not what it does to your strict press.
The nearest evidence points the other way, though it doesn't test the push press either. Pallarés 2021 in the Scandinavian Journal of Medicine and Science in Sports — 16 studies, PROSPERO-registered, GRADE-assessed — found full range of motion produced significantly greater strength gains than partial range, ES = 0.56, p = 0.004. Push press isn't a partial-range lift, so that isn't a direct hit. What it does say is that "spend more time in the top half" has never been the thing that made a lift go up in a trial, and the tip rests on exactly that idea.
What survives is smaller than what I claimed. Push press is a fine way to accumulate overhead volume. It is not a strict-press builder, and calling it "huge" was me repeating something I'd heard. Of my eight tips, it belongs last.
Do your triceps or your upper back limit your overhead press?
Your triceps limit the top of the press and your shoulders limit the bottom — and of the two accessory arguments in my reel, only the triceps one has a documented mechanism.
Back to Evangelista 2025. In the simulation, elbow extensors contributed minimally early in the lift but became dominant near full extension, and adding elbow torque shifted the velocity minimum from 38 cm to 23 cm above the chest. The authors conclude you should target shoulder strength early and elbow strength near lockout. The honest limit: it's a validated simulation, not a training trial, and no study shows close-grip bench or skull crushers raise anyone's press 1RM. Mechanism documented, carryover assumed — train them anyway, because the case for bigger triceps doesn't rest on this lift.
The upper back is the third walk-back. No trial shows that rear delt work, face pulls, or trap work increases press 1RM, and I implied one does. Kolber's 2010 review in the Journal of Strength and Conditioning Research reports that up to 36% of documented resistance-training injuries occur at the shoulder complex, naming joint and muscle imbalances plus poor technique as trends that raise risk — but it also says a majority of that research was retrospective, with a paucity of prospective work. The imbalance idea is described there, not tested.
What is well tested is lifting in general. Lauersen's 2018 meta-analysis in the British Journal of Sports Medicine pooled 6 RCTs, 7,738 participants aged 12-40 and 177 injuries, and found strength training cut injury risk with a relative risk of 0.338 (95% CI 0.238-0.480) — with a 10% increase in strength training volume reducing risk by more than four percentage points. Those were sports injuries, not shoulder injuries in lifters, so it's an argument for doing the work at all, not rear delts versus no rear delts.
One aside from Saeterbakken: posterior deltoid activation ran about 25% lower seated than standing with a barbell (p < 0.001) and 24% lower with dumbbells (p = 0.002), so pressing standing already recruits the rear delt more than you assume. Keep the upper-back work for the injury volume and the size — just don't expect it to add plates.
Does bracing and controlling the descent actually matter?
Yes — but for measurement, not for the reasons I gave in the reel.
Start with what doesn't exist: there is no trial on glute squeeze or core bracing and overhead press performance or injury. The defensible reason is that hyperextending turns the press into a different exercise. Flare the rib cage and the bar path shifts forward, the leverage changes, and this week's number stops being comparable to last week's. Hold the position constant and the number becomes a measurement you can act on.
The descent is where I overreached. I said sloppy descents beat up your shoulders. Schoenfeld 2015 in Sports Medicine pooled 8 studies and found hypertrophic outcomes similar across repetition durations from 0.5 to 8 seconds, with only volitionally very slow reps beyond 10 seconds looking inferior — and the authors flag a lack of controlled studies even there. On the injury claim, no trial shows a controlled eccentric protects the shoulder, and I don't have data of my own that separates tempo from everything else I do. Control the descent because a crashed rep puts you in a different start position for the next one. That reason holds.
What does a week of overhead press training actually look like?
Two pressing days inside a four-day week, two to three hard sets of 5-8 reps on the press, everything stopped at 1-2 reps in reserve.
I lift four days a week on a Push/Pull/Legs/Upper split around a full-time software engineering job. I'm 37, and I went from 285 lbs to about 200 at 8-10% body fat, held year-round. Four days isn't magic. It's the number that survives a real calendar.
Here's the layout I teach, not a copy of my logbook. Push day opens with the strict overhead press: 2-3 sets of 5-8 reps, all at 1-2 RIR. Press it first, when you're fresh. Upper day carries a second pressing exposure — lighter, fewer sets, same rule. Triceps work goes on the pressing days, after the press. Upper-back work — face pulls, band pull-aparts, rows — goes on pull day, for the injury volume rather than the bar. Push press is optional, once a week at most: extra overhead volume, not a strict-press builder.
The 5-8 isn't arbitrary. Lopez 2021 in Medicine and Science in Sports and Exercise network-meta-analyzed 28 studies and 747 adults, every set taken to failure, and found strength gains superior on high loads (8 reps or fewer) over low (more than 15 reps): SMD 0.60-0.63, p < 0.001-0.003. Hypertrophy was load-independent. Two honest limits: high load only edged moderate (9-15 reps) by SMD 0.26-0.28, which missed significance (p = 0.068), and nothing there separates five reps from eight — same bucket. Pick the end you can keep your setup identical on.
The progression rule is the whole system: add about 5 lb only when every work set hits the top of its rep range at 1-2 RIR. Not when the weight feels light — when the log says every set cleared it. Instrument, observe, adjust, applied to one bar. If you want the week around it, this is the 4-day split and the progression rule as one system.
One last thing: a bigger press and bigger-looking shoulders are different jobs. Width comes from side delt size, and a heavier press doesn't hand you that on its own — which is why building bigger shoulders runs on lateral raise volume while this one runs on load and where you stop your sets.