How to Track Personal Records in the Gym

What counts as a personal record

Add 5 kg to your bench press and you have set a record. Do last week's weight for two extra reps and you have also set a record, on a different measure. Both are progress. Only one of them is the kind most apps put a badge on.

Every exercise you log produces four records, not one:

Two of those need a note. A rep record is tied to a specific load, so eight reps at 100 kg and eight reps at 90 kg are separate records rather than one. Volume is simply weight times reps, so a set of 60 kg for 20 comes to 1200 and beats 100 kg for 5, which comes to 500.

For an exercise you time instead of counting, such as a plank or a dead hang, the rep record becomes a duration record: your longest hold at that load.

Most lifters say "PR" and mean the first row of the table. That is also the record that moves least often, which is why someone who counts only weight records can train productively for two months and conclude that nothing happened.

Which records actually mean something

The four are not equally informative. From most to least:

Reps at a fixed weight is the clearest signal you can get. The load is identical to last time, so the only thing that changed is how much of it you could do. Nothing else needs holding constant, because nothing else moved. That is why double progression is built on this record, and double progression for intermediate lifters covers running it as a rule rather than noticing it afterwards.

Heaviest weight matters, but only against a similar rep count. Adding 5 kg while dropping from eight reps to two is not obviously better. It is a different set. Judge it against your previous best at a similar rep count, or convert both to an estimate.

Estimated one-rep max folds weight and reps into one number, which is exactly why people like it, and it inherits the formula's error along the way. Prediction equations disagree with each other and with a tested maximum, and they get less accurate as the rep count rises (LeSuer et al., Journal of Strength and Conditioning Research, 1997). A new estimate off a set of 10 is weaker evidence than the same jump off a set of 4.

If your app also folds a logged RPE into the estimate, a second source of error comes with it, because people judge their proximity to failure imperfectly and tend to underpredict how many reps they had left (Halperin et al., Sports Medicine, 2021). So treat a 1 kg improvement as noise and a repeated upward drift as signal. How to track progressive overload goes further into the formula and its limits.

Volume in a set is the easiest of the four to set by accident. Weight times reps rewards light, high-rep work, so an easy set of 20 can beat a hard set of 5 that did more for your strength. Worth logging, worth being suspicious of on its own.

None of that makes the weaker records useless. It means you base decisions on the first two and read the other two as context. For where records sit inside the wider principle, see the complete guide to progressive overload.

One session, three records

Say your best set on an exercise is 100 kg for 5 reps, and it has been for a few weeks. Using the Brzycki formula, 1RM = weight / (1.0278 - 0.0278 x reps):

The middle row sets three records without the bar loading changing at all. If your app reports only the heaviest weight, that session looks like nothing happened.

The bottom row sets a weight record while the estimate falls. 111.2 kg is below the 120.0 kg from the week before, so on this formula the heavier set is the weaker performance. That is not an argument against heavy triples. It is a reason not to read a weight record as proof you got stronger.

The bottom row also sets no rep record, even though you had never done three reps at 105 kg before. The first time you use a weight there is nothing to beat, so that set becomes the baseline for 105 kg instead. Your first session on a new exercise works the same way: it sets the marks, it does not break them.

The rep max table

After a few months of history, the more useful view is not a list of records at all. It is your best weight at each rep count, side by side with what your estimate says you should manage there.

Continuing the example, with a best estimated one-rep max of 120 kg:

Read the gaps rather than the numbers. The 7 rep row matches its prediction exactly, because that is the set the estimate came from. The 3 and 4 rep rows sit 8 kg or so under prediction, and the honest explanation is nearly always that you rarely train there, not that you are weak at low reps. The 12 rep row is 13 kg light for the same reason.

That gives you two things. When you are about to train in a rep range you have been avoiding, the predicted column is a more grounded starting point than picking a percentage out of the air. And a row you have filled in recently, in a range you do train, that still sits far below prediction is worth a look at your technical standard or your rest periods.

You can work out a single row of this yourself with the one-rep max calculator, which turns one set into an estimate and the training weights that follow from it.

Records without a barbell

Records still work when there is no load to add.

On bodyweight exercises the rep record is the whole thing: most push-ups in a set, most pull-ups before your form goes. There is no weight to separate them by, so each rep record stands against your all-time best on that exercise.

On timed work the record is duration. A 75 second plank beats a 70 second plank. Progress is measured in seconds, and it is still progress even though nothing on the screen says kilograms.

On weighted calisthenics the estimate only means something if your bodyweight is part of it. A 20 kg belt on an 80 kg lifter is a 100 kg pull-up, and an estimate built on the 20 kg alone is not describing anything real. Log your bodyweight if you want that number to be worth reading.

Progressive overload without adding weight covers the other variables you can move when the load is fixed.

When chasing records goes wrong

Records are a scoreboard, and scoreboards change behaviour. Four failure modes are common enough to plan around.

Technique drifting to protect the number. Half a rep at the bottom, or a bounce off the chest, can produce records for months while the actual training stimulus falls. Pick a technical standard and count reps against it, or your log ends up measuring your standards rather than your strength.

Testing a true one-rep max too often. A tested maximum is hard to recover from and it interrupts the training that would have raised it. The estimate exists so that you do not have to.

Expecting a record every session. Past the beginner phase, most sessions set nothing. That is what normal training looks like. Sleep, food, stress, and how the last session went all move performance day to day.

Reading records on light accessories as progress. A rep record on a lateral raise is fine. It is not evidence about your press.

If nothing has moved on your main lifts across four to six weeks of comparable sessions, that is a pattern rather than a bad day. Why your lifts stopped progressing works through the causes in the order worth checking.

What to expect from an app that tracks records

Most workout trackers record PRs. They differ in ways that only start to matter once you have a year of history, so there are five questions worth asking of whichever one you use.

1. Does it compare against your whole history, or only your last session? A record you set in March should still be the mark in September.
2. Does it keep the four record types apart, or collapse them into one badge? A single "PR" cannot tell you whether you added weight or added reps, and those lead to different decisions.
3. Does it count your first session as records? It should not. Every set on day one is a baseline.
4. Does it tie rep records to the weight you used? "Most reps ever" on an exercise you also train light is a meaningless number.
5. Does it keep records per exercise and equipment? A dumbbell press and a machine press are not interchangeable, and neither are their records.

Lift, which publishes this page, answers them in that order. Records compare each set against your full history on that exercise, not just the last session. All four types are tracked separately, with a duration record for timed exercises. A first session sets baselines rather than records. Rep records are tied to the weight you used. Records are kept per exercise, so a dumbbell press and a machine press keep their own.

All of that sits in the free tier. Estimated one-rep max uses Brzycki, adjusted by the RPE you log, so recording effort changes the estimate. On weighted exercises the exercise screen also shows the rep max table above, for 1 to 12 reps. Full progress charts, and the weekly and monthly review that summarises records over a period, need Premium; seeing your records does not. The free versus Premium breakdown lists both tiers in full.

Other trackers do this too, and some do parts of it differently rather than worse. If you are still choosing, seven progressive overload apps compared covers the field, or see how Lift handles progression.

The point of a record is not the badge. It is that four numbers moving at different rates describe what your training is doing far better than one number that only changes when you add a plate.