Deep dive · about 9 min
Romanian deadlift: how to log it, load it and progress it
Three sets of 8 to 12 at RPE 8, 2.5 kg at a time, and a history the deadlift never touches. Here is what 90 kg for ten is worth, the ramp before it, and the exact reps that move the bar.
A Romanian deadlift starts at the top. The bar begins at the hips, slides down the thighs with the knees slightly bent until the hamstrings stop it below the knee, and comes back up the same line. The floor never enters into it, which separates the lift from the deadlift and is why it keeps its own history in the app. For a 75 kg lifter working at 90 kg, ten reps at RPE 8 is worth an estimated 132.4 kg for one, computed with the model behind the one-rep max calculator. Run for three sets of 8 to 12 on double progression, the next 2.5 kg arrives once all three sets reach twelve.
The numbers below are compute.mjs output, the script that runs the app's formulas, or carry a citation to the study they came from.
What the Romanian deadlift trains
In the app's exercise library the Romanian deadlift is a compound hinge: hamstrings and glutes primary, spinal erectors secondary. A systematic review of 19 deadlift studies found the Romanian deadlift in two of them, and its practical advice was that variants with the knees held near extension favour the back of the thigh, while the conventional deadlift is the choice for the front of the thigh and the lower back [1].
The studies that measured it agree. Twelve weight-trained men did singles at 85 per cent of their max on the leg curl, good morning, glute-ham raise and Romanian deadlift, and hamstring activity peaked in the Romanian deadlift and the glute-ham raise [2]. The semitendinosus, the inner hamstring, was far more active than the biceps femoris on every exercise [2].
Ten competitive bodybuilders lifting 80 per cent of their max for six reps showed the same split between the two hinges. The Romanian deadlift beat the stiff-leg version on semitendinosus activity during the lift, an effect size of 1.38, and the stiff-leg deadlift produced more gluteus maximus activity on the way up, an effect size of 0.99 [3].
How to log it
A Romanian deadlift is a load lift, not per-side and not bodyweight plus load, so one set is one line with the weight on the bar. The sets on the way up are warm-ups and never count toward volume. On an 8 to 12 rep hinge all three working sets are normal sets at the same load, there is no top set, and the app reads all three when it decides the next weight.
Rate every working set by reps in reserve: RPE 10 means no reps left, 9 means one and 8 means two [4]. A set logged without RPE is treated as RPE 10, the assumption every classic formula makes, so 90 kg for ten with the rating blank estimates 121.8 kg instead of 132.4 kg. Double progression reads the rating when it judges whether the ceiling was reached honestly, so the gap costs more here than on a heavy pull.
Run four warm-ups and three working sets through the app's volume code and the hamstrings and the glutes each get three sets. The spinal erectors are a secondary muscle on the lift but not one of the 14 muscles the app sets weekly targets for, so the sets credit nothing to a lower-back total. The warm-ups add nothing anywhere, as the post on whether warm-up sets count as volume explains.
The warm-up ramp for 90 kg
The app builds the ramp from the planned weight of the first working set. For 90 kg, compute.mjs warmup --kg 90 returns four sets. The 90 per cent single only switches on above 120 kg.
| Set | Load | Reps | Effort | Plates per side |
|---|---|---|---|---|
| Bar | 20 kg | 8 | about RPE 2 | none |
| 50% | 45 kg | 5 | about RPE 3 | 10, 2.5 |
| 70% | 62.5 kg | 3 | about RPE 4 | 20, 1.25 |
| 85% | 77.5 kg | 2 | about RPE 5 | 25, 2.5, 1.25 |
The working weight is a 25 and a 10 on each side, from the plate calculator. The 77.5 kg double is the one to keep: it shows where the hamstrings stop the bar before the first set of ten. A lifter who pulled from the floor earlier in the session can skip the bar and the 45 kg set.
What 90 kg for ten at RPE 8 is worth
Ten reps with two in reserve means 90 kg was a 12-rep load that day. The model reads the reps-by-RPE chart along its diagonal, reps plus reps in reserve, and lands on 68 per cent of 1RM: an estimated max of 132.4 kg. Epley and Brzycki both say 120 kg, because both treat the tenth rep as the last one available. The estimated 1RM guide covers how the chart is read.
Nobody should test that single. The number is a strength reading from session to session, not a target. The ladder below, from compute.mjs ladder --kg 90 --reps 10 --rpe 8, is the heaviest load for each rep count at RPE 10 on that day.
| Reps | Load at RPE 10 | Per cent of 1RM |
|---|---|---|
| 1 | 132.4 kg | 100 |
| 3 | 122.1 kg | 92.2 |
| 5 | 114.3 kg | 86.3 |
| 8 | 104.1 kg | 78.6 |
| 10 | 97.8 kg | 73.9 |
| 12 | 90 kg | 68 |
Read the ladder at RPE 8 instead and every load drops two rungs: 97.8 kg for eight, 90 kg for ten and 85.9 kg for twelve, computed with the same model. Twelve at RPE 8 on 90 kg is worth 138.7 kg against the 132.4 kg the ten showed, and that gap is what the next few weeks have to earn before the load moves.
The range is a choice the research leaves open. A meta-analysis of 21 studies with every set taken to failure found muscle growth similar above and below 60 per cent of 1RM, while one-rep max strength favoured the heavier loads [5]. On the ladder, 8 to 12 reps spans 68 to 78.6 per cent of the estimated max, above that line at every rung.
How double progression moves the bar
The app has seven progression schemes: Auto, Double, Top set + back-off, Load, Reps, RIR and Manual. For three sets of 8 to 12, Double fits, and under Auto a range that wide runs as Double anyway. The rule holds the load, chases one more rep each session, and adds the step once the sets reach the ceiling. For a barbell lift the step is 2.5 kg, computed with the app's model. The post on double progression covers the rule in full. Here is how it reads a Romanian deadlift at 90 kg with the default trigger, every set at the ceiling.
| Logged working sets at RPE 8 | What the rule writes next |
|---|---|
| 90 kg x 9, 8, 8 | Keep 90 kg, chase 10 reps |
| 90 kg x 12, 11, 10 | Keep 90 kg, chase 12 reps |
| 90 kg x 12, 12, 12 | Add 2.5 kg to 92.5 kg, reset to 8 reps |
| 90 kg x 14, 14, 14 | Jump to 95 kg, reset to 8 reps |
| 90 kg x 7, 7, 6, first time at this load | Hold 90 kg, could be an off day |
| 90 kg x 7, 7, 6, again | Ease to 87.5 kg, target 12 reps |
The second row is where the trigger shows: the best set reached twelve, so a best-set trigger would have added the 2.5 kg there, and the default waits for the other two sets. The fourth row is the rule noticing the load was too light, so it takes two steps and resets.
The first session at 92.5 kg will read lower on the estimate, and that is the model doing its job. 92.5 kg for eight at RPE 8 is worth 125.2 kg against 138.7 kg for the twelve that earned it, because the model rates twelve reps with two in reserve above eight with two in reserve, whatever the load. Like with like, 90 kg for eight at RPE 8 was worth 121.8 kg at the start of the block and 92.5 kg for eight is worth 125.2 kg.
A planned deload week never emits an increase, and a load increase never leaks through on a cut. More than 30 days off resets the lift: the same 90 kg for ten after 35 days away comes back as an 85 kg suggestion.
Where lifters stall on the Romanian deadlift
The usual stall is a rating that drifts. Twelve reps at RPE 10 on 90 kg is worth 132.4 kg, exactly what ten at RPE 8 was worth, so a ceiling reached by grinding earns a load step the lifter does not yet have the strength for. The twelfth rep comes with two in reserve or it is not the twelfth rep.
Another is letting the knees bend more as the weight climbs. The bodybuilders in the three-variant study did the Romanian deadlift with the knees slightly bent and the stiff-leg version with them extended, and the two gave different hamstring and glute readings at the same percentage of max [3]. Let the knees bend a little more each week and the hinge drifts toward a squat, more load moves, and the log records a Romanian deadlift that was not one.
A third is cutting the range short to keep a number. The same authors put the extra activation of their step variation down to a longer range of motion, although they did not measure it [3]. The bar goes as low as the hamstrings allow with a flat back.
What changes when you swap it?
A stiff-leg deadlift is the nearest swap: same bar, same path, knees held straight. Expect more glute and less inner-hamstring work at the same load [3]. A Romanian deadlift standing on a step keeps the knee bend and lengthens the range, and it raised gluteus maximus, semitendinosus and longissimus activity on the way up with effect sizes from 0.82 to 2.12 in a sample of ten, a direction rather than a promise [3].
A dumbbell Romanian deadlift changes the step. The app's increment for a dumbbell compound lift is 2 kg, computed with the app's model, and a rack that steps by more than that leaves the rule adding reps for longer between loads. A conventional deadlift is the wrong swap for the job: the systematic review puts the erector spinae and the quadriceps ahead of the glutes and the biceps femoris across deadlift studies [1]. Whatever you swap in starts its own history, and the 132.4 kg estimate stays with the lift that produced it.
Questions
Should I log the Romanian deadlift separately from the deadlift?
Yes. Pick Romanian Deadlift from the library rather than logging it under the deadlift at a lighter weight. Each exercise keeps its own working sets, estimated max and next-weight suggestion, so a 90 kg Romanian deadlift never drags a heavy pull's estimate down and the pull never pushes the hinge's next weight up. One lift, one history.
How much weight should I add to a Romanian deadlift?
Add 2.5 kg, and only once every working set reaches the top of the range at the RPE you set. For three sets of 8 to 12 at RPE 8 on 90 kg, that means 12, 12, 12 with two reps in reserve on each, and the next session is 92.5 kg for 8. If every set beats the ceiling by two or more reps, the app jumps 5 kg instead.
Sources
- Coratella G, Tornatore G, Longo S, Esposito F, Cè E (2022). An electromyographic analysis of Romanian, step-Romanian, and stiff-leg deadlift: implication for resistance training. International Journal of Environmental Research and Public Health 19(3):1903. doi:10.3390/ijerph19031903
- Martín-Fuentes I, Oliva-Lozano JM, Muyor JM (2020). Electromyographic activity in deadlift exercise and its variants. A systematic review. PLOS ONE 15(2):e0229507. doi:10.1371/journal.pone.0229507
- McAllister MJ, Hammond KG, Schilling BK, Ferreria LC, Reed JP, Weiss LW (2014). Muscle activation during various hamstring exercises. Journal of Strength and Conditioning Research 28(6):1573-1580. doi:10.1519/JSC.0000000000000302
- Schoenfeld BJ, Grgic J, Ogborn D, Krieger JW (2017). Strength and hypertrophy adaptations between low- vs. high-load resistance training: a systematic review and meta-analysis. Journal of Strength and Conditioning Research 31(12):3508-3523. doi:10.1519/JSC.0000000000002200
- Zourdos MC, Klemp A, Dolan C, et al. (2016). Novel resistance training-specific rating of perceived exertion scale measuring repetitions in reserve. Journal of Strength and Conditioning Research 30(1):267-275. doi:10.1519/JSC.0000000000001049