Shoulder Press Pyramid Calculator

A pyramid is a run of sets where the weight rises or falls and the reps move the other way. Heavy sets get few reps, and light sets get more^[1].

• Ascending. Light to heavy. Your heaviest set comes last.
• Descending. Heavy to light. Your heaviest set comes first.
• Triangle. Up to the heaviest rung once, then back down through the lighter ones.

Each step of the ladder is a rung. A triangle plays every rung below the top twice, so it holds more reps than the same rungs done once.

Names vary between sources. Cattan writes light-to-heavy and heavy-to-light, and uses "reversed pyramid" for a different scheme, where reps rise at a fixed weight before a light-to-heavy phase^[1]. This page says ascending and descending to keep that clear.

A 2023 review of 15 studies found pyramid training was not better than traditional training for acute responses, strength gains or muscle growth^[2]. Nine of the studies were longitudinal, and the review rated the studies good to excellent in quality^[2].

The authors say the choice can come down to periodization, motivation or personal preference^[2]. They add that this rests on studies using 8 to 12 reps and/or 67 to 85% of 1RM. Very heavy, low-rep pyramids fall outside that range.

The three abstracts report chest press, knee extension, preacher curl, leg press and leg extension results. None names a shoulder or overhead press, and all three used ascending pyramids, so none speaks to descending or triangle versions.

A 2026 narrative review mapped 40 studies from 2021 to April 2026 on methods such as drop sets, rest-pause and pyramidal loading^[6]. Several found no significant difference in strength or hypertrophy outcomes, but small samples and the lack of equivalence tests rule out concluding the methods are equal^[6].

Individual studies also reported context-specific differences, such as in session efficiency and fatigue distribution. The authors suggest adaptations are governed mainly by load, volume, frequency, effort, density and exercise selection, and they call their findings descriptive and exploratory^[6].

The 2026 ACSM position stand pooled 137 systematic reviews^[7]. It found strength was enhanced by heavier loads (80% of 1RM or more) and by 2 to 3 sets, and it lists set structure among the variables that did not consistently change outcomes^[7].

Its set structure category covers drop, cluster, complex, contrast and rest redistribution sets^[7]. The stand does not mention pyramids, so it is not a verdict on them.

Which shape? Cattan's narrative review calls the strength comparison between ascending and descending pyramids equivocal. It also says increasing, decreasing and triangle pyramids affect strength, hypertrophy and power similarly, but only triangle pyramids are likely to improve endurance^[1].

In theory, you're more fatigued on the heavy sets of an ascending pyramid. In practice, a descending pyramid needs an extensive warm-up before the heavy set^[1].

Reps and effort. The calculator reads a cell from Table 2 of Helms and colleagues. It gives the percentage of 1RM for 1 to 8 reps at RPE 7 to 10^[8]. On this scale RPE 10 means no reps left, and RPE 8 means about 2 reps left^[8][9].

Only three cells are measured values, taken from experienced squatters in the Zourdos study: 1 rep at RPE 10 (100%), 1 rep at RPE 8 (91%) and 8 reps at RPE 7 (70%)^[8]. Every other cell is an estimate.

Helms and colleagues warn that the chart is based on the barbell back squat, and the relationship "may change with machine-based, single-joint, or upper-body exercises"^[8]. The shoulder press is an upper-body lift.

They also show how much lifters differ. For 8 reps at 70%, the standard deviation was 1.2 reps in reserve, so about two-thirds of lifters may rate that set anywhere from 2 to 4 reps left. The authors say the chart should not be used as an absolute conversion tool^[8].

Your own percentage. The planned reps are compared with a 2024 meta-regression of 269 studies on how many reps people do to failure at each percentage of 1RM^[10]. The table below is its main model.

The authors made separate tables for the bench press and the leg press, and say the main table is most applicable for all other exercises^[10]. They list the overhead press among exercises with minimal data, so there is no shoulder press table^[10].

These rows match the table printed in the paper's Figure 2, which also lists 15%, 20% and 25%. For a percentage between two rows, the check compares your reps with the rows on either side, because reps to failure fall as the percentage rises. It does not estimate a value in between.

Percentages are of your own max for that lift. A press at 70% is 70% of your press max, not of your squat or bench. What differs between lifts is how many reps that percentage allows.

The meta-regression found about 13.1 reps for the leg press at 80% of 1RM against 8.8 for the bench press^[10]. In a study of trained and untrained men, the back squat allowed more reps than the bench press or arm curl at 60% of 1RM, with smaller gaps at 80 and 90%^[11].

Hold the effort the same and the percentage rises as the reps fall, so every set feels equally hard. To turn the same chart into a 1RM, use the RPE calculator.

Bold cells are measured values from the Zourdos study. Every other cell is an estimate^[8].

The next two tables are ascending pyramids of 8, 6 and 4 reps at RPE 8, which is 73%, 79% and 83% of 1RM. Loads are rounded to the nearest 2.5 kg or 5 lb. They show the arithmetic and are not a recommended plan.

Reverse the row for a descending pyramid. For a triangle, repeat the 8 and 6 rep sets after the top set.

Your shoulder press 1RM is the heaviest weight you can press from your shoulders to lockout for one rep. Overhead press, shoulder press, military press and strict press get used loosely. Kroell and Mike note that USA Weightlifting and the NSCA define the military press with conflicting foot positions^[13].

This page means a strict standing press with no leg drive. A push press is a different lift^[13], so don't use it to set your max.

For dumbbells, enter one dumbbell and read every load as per hand. The arithmetic is the same. The sources cited here don't test the estimate equations or the effort chart on dumbbell, seated or machine presses, so those results are rougher.

A tested single is the cleanest input. If you haven't tested one, the calculator can estimate a max from a set, using seven published equations for a set to failure^[14], or the effort chart if the set stopped short^[8].

Those equations weren't built on the press. LeSuer and colleagues tested them on the bench press, squat and deadlift, in 67 untrained college students^[14].

A conference abstract tested seven equations on a press, in 30 resistance-trained people. It doesn't say which press. Three equations read significantly low from sets at 70% of 1RM, and four read significantly high at 90%^[15].

The AIS testing manual is stricter. It says predictive equations should not be used to calculate a 1RM, and cites large variability (±10%) when submaximal reps are used to predict maximal strength^[16]. It asks testers to record the weight actually lifted and the reps completed.

Effort ratings aren't exact either. A 2022 review of 12 studies (414 people) found people under-predicted how many reps they had left by 0.95 on average, with large differences between studies^[17]. Predictions were slightly better closer to failure, and the review found no clear difference between upper-body and lower-body exercises^[17].

For a fuller max estimate, use the overhead press 1RM calculator. To see the plates for each set, try the barbell plate calculator.

For warm-ups and rests, see the warm-up calculator and the rest time calculator. The bench press pyramid calculator and the squat pyramid calculator do the same job for those lifts.

• Warm-up sets and rest times. The sources cited here give no standard protocol for either in a pyramid.
• Goal presets, experience levels and a diamond pyramid. The sources cited here do not tie a pyramid to a goal or an experience level, and Cattan names no diamond pyramid^[1].
• Effort chart range. It covers 1 to 8 reps at RPE 7 to 10 and is not extended. For more reps, use your own percentages.
• Shoulder press data. The chart comes from squatters and may change with upper-body exercises^[8]. The rep check's source lists the overhead press among exercises with minimal data^[10]. The sources cited here do not cover dumbbell, seated or machine presses.
• Fatigue between sets. The rep table uses only first sets, done in a non-fatigued state^[10]. In studies of repeated sets to failure at one load, set 2 averaged about 70% of the reps in set 1, and set 3 about 55%^[12]. Those sets used one load and went to failure, so the figures do not carry over to a pyramid, and this page does not model the effect of earlier sets.
• Who the rep data describe. Most samples were healthy (97%), under 59 years old (92%) and male (66%), with a median sample age of 23^[10]. The authors found no clear effect of sex, age or training status, and say the impact of most moderators was uncertain.
• How the rep data were gathered. The studies used tested 1RMs, so an estimated max adds its own error to the rep check. The authors say their search did not follow standard meta-analysis guidelines^[10].
• Abstract-only sources. The three pyramid trials, the 2023 and 2026 reviews, Zourdos, Shimano, LeSuer, Halperin, the repeated-sets analysis and the Varner conference abstract are cited from their abstracts only.
• Technique, recovery and progression. This page gives loads, reps and totals. It does not cover technique, recovery or how to progress a pyramid over time.