Percentage of 1RM for power: ACSM 2026 widens the band to 30–70%

For roughly seventeen years, “power training” meant 50–75% of 1RM — the band the 2009 ACSM Position Stand on Progression Models in Resistance Training framed as moderate load with intent. The 2026 update widens that to 30–70% of 1RM and explicitly endorses Olympic-style weightlifting and power resistance training alongside it. The force-velocity curve has two productive ends, and the older band only honored one.

That widening matters at the gym floor: a 30% jump-squat with an empty barbell, a med-ball chest pass, a loaded countermovement jump — these all develop power even though the older band would have flagged them as too light. To put real numbers behind the change, run a 1RM through the percentage-of-1RM calculator, and check the dual-framing table on the methodology page for how the NSCA traditional and ACSM 2026 framings sit side by side.

The power band, before and after ACSM 2026

NSCA's programming-percentage frame puts power at 50–75% of 1RM, with low rep counts, long rest, and a fast concentric. That frame is the field default and lines up with the older 2009 ACSM Position Stand (Ratamess et al., Med Sci Sports Exerc 41(3):687–708, 2009), which described power training around moderate loads in roughly the same range. The NSCA band is what the StrengthMath load table highlights by default — most written programs over the last three decades were calibrated to it.

ACSM 2026 (Currier et al., Med Sci Sports Exerc 58(4):851–872, April 2026) widens the band to 30–70% of 1RM. The relevant line from the abstract, verbatim: “Power was enhanced by moderate loads (30%–70% one repetition maximum), low-to-moderate volume (≤24 repetitions⋅sets), Olympic-style weightlifting, and power RT.” That single sentence is the entire shift this article rests on.

The word that did the work between 2009 and 2026 is moderate. In 2009, “moderate” meant the middle of the loading spectrum — 50–75% — and ballistic work in the 30–50% range was treated as a separate speed/velocity category. The 2026 review redefines moderate as anything 30–70%, which absorbs the old speed-day range into the power band. My read: this isn't a new finding so much as a recategorization. Light-load fast work was always developing power; the 2009 framing just put it in a different column.

Why the force-velocity curve matters

Power equals force × velocity. At 30–50% of 1RM the bar moves fast, velocity is the dominant term, and power output rides the velocity end of the curve. At 60–70% the bar moves slower but with much more force, and power output rides the force end. The curve has a peak — typically somewhere between 30% and 60% depending on the lift and the lifter — but power production is meaningful across the whole 30–70% range. That's why the ACSM 2026 band is wide instead of a single optimal percentage.

The practical consequence: if you only work in the 60–75% slice, you train the force end of the curve and leave the velocity end underdeveloped. If you only work in the 30–45% slice, you train the velocity end and leave force expression flat. A program that targets power for a sport — anything where speed-strength matters — should touch both halves of the band in the same training week, not pick one and call it done.

The load table at 100 kg

To make the band concrete, here's the engine's load table for a 100 kg 1RM, with the in-band rows flagged. The engine highlights the NSCA traditional band by default (the default visual channel for the load table), and the ACSM 2026 column shows where the widened band picks up additional rows below NSCA's lower edge:

Engine output for oneRepMax: 100, goal: "power", unit: "kg". NSCA in-band rows match the engine's inBand flag (rows 50–75% inclusive). The ACSM 2026 column is annotation against the band cited in Currier et al. 2026 — it is not a separate engine channel.

The two bands overlap from 50–70% — that's the conservative core where both framings agree. NSCA picks up 75% on the heavy edge (still moderate by traditional periodization standards). ACSM 2026 picks up 30%, 35%, 40%, and 45% on the light edge, which is the velocity-emphasis range. If I had to pick one default for a general power block, I'd still center it at 50–70% — that's where the two framings agree and where the loaded movements feel like recognizable power work — and add 30–45% velocity-day work as a deliberate complement, not as a substitute.

Sets, reps, and the fast concentric

The engine's power prescription is 3–5 sets × 1–5 reps, fast concentric, 2–5 min rest, with a supplementary RIR target of 2–4. Three pieces of that line carry weight. Low rep counts (1–5) keep every rep before fatigue slows the bar. Long rest (2–5 min) lets phosphocreatine refill so rep two of set two is as fast as rep one of set one. And the fast concentric is not a stylistic note — it's the defining feature of power work.

Bar speed is the variable. A 60% squat ground out at 0.4 m/s is strength work; the same 60% squat moved at 0.8 m/s is power work. Same load, same lifter, different intent, different adaptation. If you're training power and the bar isn't moving fast, the load is the wrong load — not because 60% is the wrong percentage, but because intent collapsed. RIR 2–4 is wider than the strength RIR of 1–3 for exactly this reason: pushing to true near-failure on power day kills bar speed on the next set.

Olympic lifting, ballistic work, and where the calculator stops being useful

ACSM 2026's sentence on power names four routes — moderate loads at 30–70%, low-to-moderate volume, Olympic-style weightlifting, and power resistance training — and all four made the position stand. The percentage frame works cleanly for the first two: load a back squat or a bench press to 60% and program reps. It works less cleanly for Olympic lifts. A snatch or a clean has a moving 1RM that's technique-bounded as much as strength-bounded; doing 70% of a clean isn't the same kind of stimulus as doing 70% of a back squat, because at 70% your clean technique might still be breaking down.

For the non-competitive lifter, my take is to skip the snatch and clean unless a coach is teaching you in person — the technique cost is real. Ballistic variants of lifts you already know (jump squats, bench throws on a Smith, kettlebell swings, med-ball chest passes) deliver the velocity-end stimulus the ACSM 2026 widening was meant to capture, without asking you to learn a new sport. The calculator-and-percentage frame is reliable for those. For competitive Olympic lifting, the percentage frame still applies but the working percentages live in coach territory, not calculator territory.

What the power band does not cover

A few things the percentage frame stays silent on. Plyometrics with no external load (depth jumps, broad jumps, single-leg bounding) are power work that produces no %1RM number at all — they sit alongside loaded power work, not inside the percentage table. Sprint-speed work has its own programming logic and isn't a back-squat percentage problem. Sport-specific power expression (a swing, a throw, a tackle) is downstream of the gym number; the gym number builds the engine, the sport teaches it to fire on cue.

What the engine doesn't model and the calculator doesn't surface lives on the methodology page: age-related velocity decline, individual differences in the force-velocity peak, fatigue carryover from a prior strength block. If your power numbers stagnate inside the 30–70% band, the adjustment is usually not a different percentage — it's rest, a reload, or a check on whether the bar is actually moving fast on working sets. Trust the bar speed over the table. The table is the starting point.

Common questions

What's the best %1RM for power training?

ACSM 2026's Position Stand on Resistance Training puts power at 30–70% of 1RM, widening the older NSCA 50–75% band that most written programs are built around. The 30–50% slice emphasizes bar velocity (ballistic / jump variants); 60–70% emphasizes force production; both develop power. For a 100 kg back-squat 1RM, that's 30–70 kg of working load with a fast concentric and 2–5 min rest between sets.

Why did the power band widen from 50–75% to 30–70%?

The 2026 review (Currier et al., Med Sci Sports Exerc 58(4):851–872) synthesized 137 systematic reviews covering ~30,000 participants and reported, verbatim, that power was enhanced by moderate loads of 30–70% 1RM, low-to-moderate volume, Olympic-style weightlifting, and power resistance training. The widening reflects the force-velocity curve: light-load high-velocity work (ballistic squats, med-ball throws, loaded jumps) develops power output at the velocity end of the curve, which the older 50–75% band excluded.

Should I do Olympic lifting for power if I'm not a competitive lifter?

Not required. ACSM 2026 endorses Olympic-style weightlifting and power resistance training as effective routes — but ballistic variants of standard lifts (loaded jump squats, bench throws on a Smith machine, med-ball chest passes) deliver the same velocity-end stimulus without the technique cost. If your sport doesn't reward Olympic technique directly, the calculator-friendly path is fast-concentric work in the 30–70% band.

How does power training differ from strength training?

Strength is heavy and slow; power is anything that produces force fast. ACSM 2026 puts strength at ≥80% 1RM with 1–3 RIR; power at 30–70% 1RM with 2–4 RIR and a fast concentric. The bands overlap at the high end (60–70% trains both), but the intent differs — slow grind for strength, intentional bar speed for power. See the sister page on percentage of 1RM for strength for the heavy end.

Where to next

Power and strength share the high end of the loading spectrum — both live above 60%, and a heavy 70% squat builds both at once when bar speed is intentional. The clean separation is at the velocity end (30–50%, power-only) and the absolute-strength end (≥80%, strength-only). For the heavy half of that picture, see percentage of 1RM for strength. If you don't have a 1RM number you trust yet, best 1RM formula by lift covers which prediction equation to use per movement, and the 1RM calculator runs the math. Then drop the result into the percentage-of-1RM calculator and pick a goal.