How much stronger should you get each month?

Real numbers from Heft lifters: about 2.6% a month in your first three months on an exercise, faster on leg lifts, slower on curls, and why eight sessions without a record is normal.

Expect your estimated max on a lift to climb about 2–3% a month in your first three months on it. That’s the median across 56 exercises logged by Heft lifters.

Leg lifts climb faster and arm lifts slower. At any moment, about half of lifters are in a run of eight sessions without a new best on a given lift. Neither means your training is broken.

The short answer: In their first three months on an exercise, the typical Heft lifter’s estimated one-rep max rose 2.6% a month (median across 56 exercises). Among lifters with at least 12 sessions of an exercise, a median of 52% had gone their most recent eight sessions without beating their earlier best.

How much stronger should you get each month?

Plan for 2–3% a month on your estimated max, and more on leg exercises. Sets of eight at 60 kg on the bench press estimate a max of about 76 kg. The bench’s 2.3% a month adds about 1.7 kg to that, roughly one extra rep at 60 kg every five weeks.

Here is what the typical lifter gained per month on common lifts during their first 90 days, and how many were in an eight-session stall when we looked.

Exercise Gain per month, first 3 months Lifters In an 8-session stall Lifters
Machine Standing Calf Raise +14.8% 21 38% 16
Sled Hack Squat +11.8% 15 not enough data –
45 Degrees Leg Press Machine +11.7% 24 41% 22
Barbell Romanian Deadlift +9.9% 29 50% 16
Machine Leg Extension +6.8% 56 52% 44
Squat +3.6% 25 53% 15
Chest Press +3.6% 43 53% 30
Upper Grip Lat-Pulldown +2.5% 47 55% 38
Bench Press +2.3% 41 44% 25
Dumbbell Standing Lateral Raise +2.0% 38 68% 31
Barbell Curls +1.2% 28 43% 21
Dumbbell Hammer Curl 0.0% 31 53% 19

These rates describe the early climb, not a whole training career. The fastest rows sit on machines, where part of the early gain is learning the movement and the seat settings.

Each row rests on 15 to 56 lifters, so a single row can shift by a few points as more people train that exercise.

Why do leg lifts climb faster than arm lifts?

Compare your progress on a lift with that same lift, not with your leg press. Leg presses, the hack squat, the Romanian deadlift, and calf raises climbed 8–15% a month in the first three months; curls climbed 0–1.2%.

Powerlifting data shows the same split between lower and upper body. Christopher Latella and colleagues analysed meet results from 1,896 powerlifters between 2003 and 2018. Maximum strength gains ran 20.2–25.4 kg a year on the squat and 18.1–21.1 kg on the deadlift, against 10.5–12.8 kg on the bench press. Latella and colleagues, 2022.

Curls face a second problem: the weight steps. Moving from a 12 kg to a 14 kg dumbbell is a 17% jump, so many lifters use the same dumbbell for weeks, and the typical trend stays flat.

On curls, chase reps. One more clean rep with the same dumbbell is progress.

The powerlifting figures are in kilograms, so part of the squat’s lead comes from its heavier starting weight, and the dumbbell explanation for flat curls is a likely reason rather than something the data measures.

Is it normal to go weeks without a new personal best?

Yes. At any moment, about half of lifters have gone eight sessions of a lift without beating their earlier best, so one flat month is no reason to throw out your plan.

Across the 40 exercises with enough data, that share ranged from 36% on the cable face pull to 75% on the V-bar lat pulldown. The median was 52%.

Progress also slows with time. Alexander Renner and colleagues followed 6,524 classic powerlifters with at least ten competitions each: yearly gains generally shrank the longer lifters stayed in the sport, and lifters typically reached their best results after four to six years of competing. Renner and colleagues, 2026.

Your first months will be your fastest. A slower climb afterward is the normal shape of progress, not a sign that something went wrong.

The powerlifting study only covers competitors who kept competing for years, so it shows the shape of the curve rather than a timetable for your own lifts.

What should you do when a lift stalls?

Before you change anything, look at your last eight sessions of that lift, then change one thing at a time.

Start with reps. At 60 kg, one extra rep raises your estimated max by 2 kg, so the weight can stay the same while you get stronger.

Then check whether the sessions are comparable. A bench press after three other chest exercises isn’t the same test as one done first, and a set you stopped early says less than a set you took close to failure. The post on using your workout log to decide what comes next walks through that check.

If the weight and the reps really haven’t moved for eight sessions, pick one change: a rep target to reach before you add weight, a smaller weight step, or a close variation of the lift for a few weeks.

Each exercise page in the exercise library shows how that lift’s lifters progress, so you can check your stall against theirs.

How we measured

The numbers come from workouts that Heft lifters chose to sync, aggregated on 28 September 2026. For each lifter, Heft estimated a one-rep max from every weighted set using the Epley formula, weight × (1 + reps ÷ 30), and kept the best estimate per week.

The monthly gain is the trend across those weekly bests over the first 90 days after a lifter first logged the exercise, relative to their first two weeks. It’s the median slope across all pairs of weeks, so a single unusually good or bad week barely moves it. It counts lifters with at least six weighted sessions spanning six weeks.

A stall means the best estimate from a lifter’s most recent eight sessions was no higher than their best before that, counted among lifters with at least 12 sessions of the exercise. Every figure is a median or share across lifters and appears only when at least 15 lifters qualify. The methodology page covers the rest.

The data comes from people who chose Heft and chose to sync, holds no sex, bodyweight, or training-age information, and uses estimated rather than tested maxes, so read it as a benchmark, not a prescription.

All posts