Most apps answer this with one number. Ninety seconds, every set, every lift, forever. The others put a stopwatch on the screen and leave the question to you.
Ninety seconds is not a bad number. It is roughly the middle of the range. As a universal answer it is wrong in both directions at once: too short for a heavy squat top set, too long for a lateral raise.
How long should you rest between sets?
Between 75 seconds and 4 minutes. Which end you land on is set by two things: whether you are training for size or for strength, and whether the lift is a compound (several joints and several muscles at once, a squat or a row) or an isolation (one joint, one muscle, a curl or a lateral raise).
Two questions, four answers:
- Build Muscle, isolation: 1:15.
- Build Muscle, compound: 2:00.
- Get Stronger, isolation: 2:30.
- Get Stronger, compound: 4:00.
Why the gap between them is that wide
It is tempting to read a spread that size as rest doing the work. The evidence does not support that reading. Current ACSM guidance places rest-interval length among the variables with minimal independent effect on the outcome once volume and intensity are matched. Minimal is not nothing, but it is not the lever most people take it for. Two lifters who complete the same sets at the same loads land in much the same place whether they took two minutes or three.
The word doing the work in that sentence is complete, and completing is what the number protects. That is the reasoning the engine’s table is built on: a flat default under-recovers compound strength work, so the prescribed reps get missed, and over-rests hypertrophy isolation work, so sessions run long and feel slack. Cut a heavy squat’s rest to 90 seconds and you do not get a shorter session with the same training in it. You get a lifter who misses the last two reps, and the volume and load that drive the adaptation come off quietly, with nothing on the screen saying so.
One limit of the table is worth naming. It reads your goal and the lift type, and the rules around it read the load. None of them reads whether the bar is on your back. A coach watching a squat and a leg press at the same load would not call them the same demand, and the rest number does not tell them apart. If you are going to hold yourself to longer than the card asks anywhere, the spine-loaded compounds are the place to do it. The engine does draw that line elsewhere, in what it refuses to pair, which is further down.
The bands themselves are not Sterka’s invention. The IUSCA hypertrophy position stand puts multi-joint work at two minutes or more and single-joint or machine work at 60 to 90 seconds, and notes that longer rests are at least as good, which reverses the older advice that short rests were better for size. Older ACSM progression guidance puts advanced lifters at three to five minutes on heavy core lifts and novices at one to two. Three of the four numbers sit inside those ranges, though not centred in them: 2:00 is IUSCA’s floor for multi-joint work rather than its middle, and 4:00 comes from a row that stand wrote for advanced lifters, which is worth knowing given Sterka hands it to anyone training off the band past the start. The 2:30 for a strength isolation is a coaching-standard pick that neither stand prices at all.
The four-cell table is the last thing the engine checks
If that table were the whole rule, your first squat session on a strength goal would carry four minutes between sets. It does not.
Rest is decided per working set, not per exercise, and the goal-and-lift-type table sits at the end of the lookup rather than the start. Four situations are checked first, in this order.
One: your first session rests 90 seconds, whichever goal you picked. The cold-start row ignores your goal entirely: 90 seconds, compound or isolation. It applies while the beginner gate is open, meaning you rated yourself as starting out and you are inside your first 28 days. It also covers anyone’s very first session, whatever experience they reported, and that half graduates after one completed session rather than after a month. For everything except a Build Muscle isolation it makes rest shorter, and not because a newer lifter recovers faster. Cold-start work is capped at 75 percent of your estimated one-rep max at a target of three reps in reserve, which is submaximal, and submaximal work does not need a strength-compound rest. On a Build Muscle isolation it runs the other way, 75 seconds up to 90, because a first session is also when you are still working out how to set the thing up. The compound row is 90 rather than 120 partly for that reason and partly for a plainer one: a 30-minute full-body session cannot hold a full warm-up ramp and 120-second rests and still end in 30 minutes. At 75 percent and three reps in reserve the shorter rest costs close to nothing, and finishing beats resting perfectly through half a session.
Two: a bodyweight accessory rests like an accessory. A movement you cannot load, programmed as accessory work at 8 to 12 reps and two reps in reserve, reads the Build Muscle row even on a strength day. Without that rule a set of push-ups, which carries no load percentage for the engine to read at all, would fall through to the 240-second strength-compound row.
Three: a light card reads the Build Muscle row. Any set below 65 percent of your estimated one-rep max reads the Build Muscle row for its lift type, whatever goal you are on. That covers a deload week’s loaded work, a re-entry compound after a layoff, and the opening set of a Get Stronger volume week, where the wave’s percentage lands below the band. Four minutes between sets of a deload squat is over-rest, and it would turn the easiest week of the block into its longest.
Four: the opening sets of a strength ramp are back-offs. A Get Stronger ramp climbs across its working sets, and the sets sitting in the band’s lowest quarter are back-offs relative to the top. Those read their own row: 3:00 on compounds. On isolations the number does not move, because a strength isolation rests 2:30 either way.
The order is as load-bearing as the list. A Get Stronger compound at 60 percent rests 2:00, not 3:00, because the light-load rule is checked before the back-off rule ever runs.
What a Get Stronger block actually delivers
Two consequences follow, and the table on its own suggests the opposite of each.
On a compound loaded off the goal band, 4:00 is the normal rest, not the rare one. How many of your sets sit in that lowest quarter depends on the shape of the ramp: a three-set squat ramp delivers 3:00, 4:00, 4:00, a four-set ramp delivers 3:00, 3:00, 4:00, 4:00, and a five-set ramp delivers 3:00 and then four sets at 4:00. In every shape it is the openers that come down and the rest of the ramp that does not. Budget for that rather than for one long rest at the top.
Your main lift on a strength block is probably not loaded off that band at all. Its weights come from the four-week wave, set as fractions of your Training Max, and the Training Max is seeded at 90 percent of your estimated max. Because it is a fraction of a fraction, a wave week written as 80 percent is nearer 72 percent of your actual max, which is under the band entirely. Across a full eight-week strength cycle the anchor rests 3:00 and 2:00 about equally often, and reaches 4:00 twice: the top back-off of the intensity block’s second week, and the heavy single in its third. A secondary compound loaded off the band reaches 4:00 every session. Your heaviest lift is the one resting least, which is worth knowing before you read it as the app losing track of your goal.
Warm-ups are on their own clock
None of the four situations above governs warm-up sets, with one exception. Warm-ups carry a rest value per rung instead of one value for the ramp. A hypertrophy ramp runs up to three rungs and rests 60, 60, then 75 seconds. A strength ramp runs up to five and rests 45, 60, 75, 120, then 150. The jump at the top is the point: the last two rungs are a double at 85 percent and a single at 92 percent of your working weight, they exist to prime the nervous system rather than to add work, and going into your first real set off 45 seconds would defeat them.
Those counts move in both directions. Lighter working sets drop the lightest rungs, and a strength ramp below a per-modality weight floor falls back to the three-rung version. Coming back from a layoff adds one instead, a set of ten at 40 percent in front of the ramp. And a first session replaces the ramp altogether with a single set at 70 percent for five reps, resting 60 seconds, which is the cold-start rule reaching into the warm-up. You can see the ramp for a given lift in the warm-up sets calculator.
Supersets change the question
Paired exercises sit outside the table entirely. A superset in Sterka is not the back-to-back-with-no-rest version: the pair runs a shared round rest of about 90 seconds plus a short station move, so each lift gets its partner’s working time as recovery.
That round delivers at least two minutes to each lift, and a lift needing more than the round gives it is refused, which rules out every strength row: 2:30, 3:00 and 4:00 are all above what a round can return. So is anything at or above 85 percent of your estimated one-rep max, and on Get Stronger the ceiling is tighter still at 80 percent. A hypertrophy compound at 2:00 is covered exactly, so it clears that gate.
One refusal has nothing to do with the clock. A lift that stacks load on your spine under a free bar never pairs, whatever it weighs and whatever rest it was given. A barbell back squat and a conventional deadlift are out. Pairing manufactures time pressure, and you cannot hold a rack while you walk to a bench.
Read that as compression rather than as effort, because the line does not fall where the lift’s name suggests. A leg press or a Smith squat still pairs: the frame carries the compression. So do the Romanian and stiff-leg deadlifts, the trap-bar pull, the goblet squat and the landmine variants. Those are carved out on purpose: they are what the engine swaps in when the rule does fire, and they would be no use as a landing spot if they were barred too. A bent-over row pairs because it never loads the spine that way in the first place. The question is what is stacked on your spine, not how hard you have to brace. That is the distinction the rest table could not draw, drawn where it matters.
When you cannot take the rest
Some sessions do not have room for it. The rack is busy, you have 40 minutes, and the card in front of you adds up to more than that. This is the part most rest advice leaves out.
Keep the rest. Shed the work instead.
The instinct is the opposite: keep every set and take the time out of the gaps. That is the one move worth avoiding. It removes the load and the reps the session was built around while leaving the card looking untouched. Sterka’s own behaviour under a time cap is a fair guide to what to drop. When a session runs long the engine takes sets off before it takes exercises off, unwinding the week’s added sets first, then thinning isolations, and only dropping a lift entirely when neither is enough. Through all three it never shortens the prescribed rest, and it never lightens the bar.
On Get Stronger, protect the load before anything else. The load is the stimulus, so a lighter top set is not a shorter version of the session, it is a different one. Take the time out of the accessories, and if that is not enough, run fewer of them.
On Build Muscle, the sets are the stimulus, so cutting them comes last. Take the time out of the isolations first. The 1:15 in the table exists because a curl has little at risk, and 60 seconds there, the floor of the published band, costs you little. Sixty seconds on a compound costs you a lot. Pair antagonists next, which is what the round rest above is for, remembering that the squat will not go. Then follow the engine’s order rather than your own instinct: thin the sets before you drop the lift, because a muscle’s only direct work at a lighter dose still beats losing it from the session.
What this means between your sets
The clock on your screen is not a default nobody got round to changing. It is matched to the set you just did and the one you are about to do, and it moves when your training moves: 90 seconds through your first session, shorter on a back-off, shorter through a deload, and four full minutes on a band-loaded compound once you are past the start.
Which leaves you less to decide. Rest is one of the few variables where guessing costs real training, because the cost does not show up as a missed timer. It shows up two sets later as reps you did not get.
References
- American College of Sports Medicine position stand. Resistance training prescription for muscle function, hypertrophy, and physical performance in healthy adults: An overview of reviews. Medicine & Science in Sports & Exercise, 2026.
- Schoenfeld BJ, Fisher JP, Grgic J, et al. Resistance training recommendations to maximize muscle hypertrophy in an athletic population: position stand of the IUSCA. International Journal of Strength and Conditioning, 2021;1(1).
- American College of Sports Medicine position stand. Progression models in resistance training for healthy adults. Medicine & Science in Sports & Exercise, 2009;41(3):687–708.