Linear progression: how far it goes and what comes after

Linear progression means adding the same small load to the bar every session, as long as you hit every rep you planned. On a barbell lift the app's step is 2.5 kg, so a squat that starts at 60 kg reaches 100 kg in 16 sessions. The run is over when the bar keeps getting heavier but the estimated 1RM stops climbing with it, and the logged RPE tells you that a session or two before you miss a rep. After that, the lift moves to double progression: a rep range at a lighter load, with weight added only when every set reaches the top.

The figures below come from compute.mjs, the script that runs the app's own formulas, or carry a citation. You can check any of the estimates against the one-rep max calculator.

How far 2.5 kg a session goes

The arithmetic is the appeal. A barbell lift in the app moves in 2.5 kg steps, a dumbbell lift in 2 kg steps, and a machine or bodyweight lift in 1 kg steps. Squat three times a week from 60 kg and the 100 kg session arrives in under six weeks, with no decisions to make in between.

That pace is only possible for a lifter who is new to the lift. A meta-analysis of 18 studies found that improvements in 1RM were greater among untrained participants than trained ones [1]. Early sessions add strength faster than the load step does, so the step feels easy for weeks. The gap closes as the lifter gets stronger, but the step on the bar is still 2.5 kg at 100 kg.

The American College of Sports Medicine's position stand puts the general rule in the same terms. It recommends a 2 to 10 per cent increase in load when a lifter can do the current workload for one or two reps over the target [2]. Linear progression is that rule applied every session, at the smallest step the gym allows.

No trial has measured how long a novice linear run lasts, so any figure in weeks is a guess. What can be measured is your own log, which says exactly when the step has outrun the strength.

When is the run over?

The run is over when the estimated 1RM goes flat while the load keeps rising. Missed reps come later. They are the last symptom to show up.

RPE makes this visible. RPE is a rating of how hard a set was on a 1 to 10 scale, where RPE 10 means no reps left in reserve, RPE 9 means one left, and so on down [3]. The app reads every set you rate through one reps by RPE model, so a set of five at RPE 7 and a set of five at RPE 9 give different estimates even at the same load.

Here is the end of a linear run on the squat for a 90 kg lifter, three sets of five, with the RPE of the last set and the estimate the model gives for it.

Read the right-hand column, not the load. Between sessions 1 and 3 the estimate rises with the bar. From session 3 to session 5 the bar gains 5 kg and the estimate moves half a kilo, from 117.1 kg to 117.6 kg. The lifter is not getting stronger any more. They are spending reps in reserve to lift the new weight, and RPE climbs from 8 to 9.5 to pay for it.

Session 6 is the miss everyone notices: four reps at RPE 10, worth 112.1 kg. Had the lifter ground out all five at RPE 10 instead, that set would be worth 115.9 kg, still less than the 95 kg session. The run ended at session 3. The log just took three more sessions to say so out loud.

So the practical signal is two things together. RPE on the last set rises by a full point or more across two sessions, and the estimated 1RM stays within a kilo or so of where it was. When you see both, stop adding load, even if you still hit every rep.

Why the run ends

Linear progression ends because the load goes up by the same amount every session and strength does not. Untrained lifters improve fastest [1], and the rate slows as they gain experience, but the step on the bar never shrinks on its own.

The same position stand that gives the 2 to 10 per cent rule separates novices from intermediates, whom it describes as having about six months of consistent training [2]. For intermediate and advanced lifters it recommends a wider loading range used in a periodised fashion, rather than one load climbing forever. That is a polite way of saying the session-to-session run is a beginner's tool.

There are other reasons a run stops early, and they are worth ruling out before you blame the program. A cut stalls loads that would otherwise keep climbing. A lifter who skips sessions loses the frequency that made the run fast. Weeks of squatting three times a week also add up to a lot of hard sets, and fatigue can flatten the estimate before strength has really stopped. If the estimate dropped rather than flattened, the guide to when a lift stalls walks through those causes in order.

What comes after: double progression

The usual next step is double progression. You hold a load, build every set through a rep range, and add weight only when every set reaches the top. Progress continues in reps between load jumps, so the bar does not have to move every session for the lifter to keep improving.

There is evidence that adding reps works about as well as adding load. In an 8-week trial, 43 lifters with at least a year of training either added load at fixed reps or added reps at a fixed load [4]. Smith machine squat strength slightly favoured the load group, by 2.0 kg, and growth of the rectus femoris, one of the quadriceps, modestly favoured the reps group. The authors called both viable strategies.

To make the handoff, set the new range from today's best estimate, not from the last load you missed. The lifter above peaked at 117.1 kg. The model says that is worth 92 kg for six reps at RPE 8 and 86.5 kg for eight reps at RPE 8. So a range of six to eight reps starts at 87.5 kg, the nearest plate-loadable weight. Six reps there should feel comfortable. When all three sets reach eight at RPE 8, that set is worth an estimated 118.4 kg, which is new strength gained without touching the load.

Before the first set at 87.5 kg, the warm-up ramp the app writes is the bar for eight, 45 kg for five, 60 kg for three and 75 kg for two. None of those sets count toward volume. The double progression guide covers the rule in full, including which sets must reach the ceiling before the load moves.

You do not have to switch every lift at once. Each lift reaches the end of its run on its own schedule, so switch each one when its own log says so.

Smaller steps can buy a few more sessions

A smaller step slows the load to match a slower rate of strength gain, and that can extend the run. On a 20 kg bar, 102.5 kg is 25 kg, 15 kg and 1.25 kg plates on each side. With 0.5 kg fractional plates, 101 kg is 25 kg, 15 kg and 0.5 kg a side, a 1 kg step instead of 2.5 kg. The plate calculator works out the loading for any target and plate set.

Smaller steps do not change the signal. If the estimate is flat across three sessions at 1 kg steps, the run is over at 1 kg steps too. They buy time, not a new run.

Running linear progression in the app

In the app, linear progression is the scheme called Load. You set a fixed rep target and an increment, and the next weight follows from what you logged last session. The program editor lets you set a smaller increment on any exercise if you own fractional plates.

What the Load scheme does with each outcome:

• Hit the target on every working set and it adds one step.
• Fall short and it repeats the same weight.
• Keep missing at that weight and it drops one step, with a warning that the jumps look too big and a suggestion to use smaller increments or double progression.
• Three sessions without progress flags a stall by default, unless the estimated 1RM is still rising.
• A planned deload week never adds load, and on a cut an increase becomes a hold.
• More than 30 days off the lift resets it before it progresses again.

If you leave a lift on Auto, the rep target decides the scheme. A fixed rep target, or a range less than two reps wide, gets Load. A range at least two reps wide gets Double. So the handoff is one edit: change the lift's target from five reps to six to eight, and Auto treats it as double progression from the next session. The guide to choosing a progression scheme explains the other schemes and when each one fits.

Mistakes that end a run early or hide its end

Logging without RPE is the commonest mistake. A set with no RPE is read as RPE 10, the assumption every classic 1RM formula makes, so every set looks like a max. The estimate then rises with the load until the first miss, and the warning signs in the table above never appear.

Resetting too far comes next. A lifter who misses at 100 kg and drops back to 80 kg spends weeks relearning weights they already own. The log already holds the best estimate, and the new range can start from that.

Some lifters treat a cut as the end of the run. Strength often stalls in a deficit, and that says more about the deficit than about whether linear progression has run its course. The app holds the load on a cut for exactly that reason.

Restarting linear progression on a lift that already finished one wastes the most time of all. A second run from a lighter weight climbs back to the same ceiling, because nothing about the rate of strength gain has changed. Moving to a scheme that progresses more slowly is the change that keeps the lift moving.

Questions

Is linear progression the same as linear periodisation?

No. Linear progression adds load every session. Linear periodisation changes the emphasis across weeks or blocks, usually from higher reps at lighter loads to fewer reps at heavier loads. A meta-analysis of 13 studies found linear and daily undulating periodisation produced similar muscle growth [5]. The two terms share a word and little else, so check which one a program means.

Should I deload before switching to double progression?

Usually not. If the estimate flattened rather than fell, the lifter is not fatigued so much as out of easy strength. Dropping to the new range at a lighter load already lowers the effort of every set. A deload makes sense when the estimate is falling across sessions or RPE stays high at loads that used to feel easy, which points to fatigue rather than the end of the run.

Can I keep linear progression on some lifts after it ends on others?

Yes. Each lift has its own log and its own ceiling, and they rarely run out in the same week. Keep adding the step to any lift whose estimated 1RM still climbs with the load. Switch each lift when its own estimate goes flat, and leave the others running until they do.