How to deadlift: conventional, sumo and Romanian

Pick a deadlift you can control, start light, and add weight once every rep looks the same. Conventional, sumo and Romanian deadlifts are all fair choices. No style has been shown to cause fewer injuries.

"Will it hurt my back?" has a three-part answer, and the parts belong together. Competitive lifters get hurt less often than contact-sport athletes. The spine is still one of the most common places they get hurt.^[1] And the idea that a rounded back causes back pain hasn't been shown in studies.^[2] Nobody has shown that heavy deadlifts with a rounded back are safe, either.

So the deadlift is neither the back-breaker of gym legend nor a lift you can't get wrong. The evidence is thinner than both camps suggest. Where we're working from coaching practice instead of studies, we say so.

Why it matters

A deadlift is any lift where you move a loaded bar between the floor (or near it) and standing, by bending at the hips. That bend is a hip hinge: you lean forward by pushing your hips back, not by squatting down. You'll meet four versions here.

• Conventional deadlift: feet fairly close together, the bar lifted from the floor.
• Sumo deadlift: a much wider stance with the toes turned out, the bar lifted from the floor.
• Romanian deadlift (RDL): you start standing with the bar. You lower it by pushing your hips back, knees a little bent, then stand up again. A stiff-leg deadlift is the same idea with straighter legs.
• Trap-bar deadlift: you stand inside a six-sided bar (a trap bar or hex bar) and lift it by its side handles.

A beginner can get the deadlift wrong in two ways. The first is skipping it out of fear for their back. That trades away a whole group of lifts over a risk that, in competitive lifters, is lower than in contact sports.^[1]

The second is loading the bar too fast. A pull from the floor makes it easy to force out ugly reps: hips shooting up first, the bar drifting away from the legs. And in competitive lifters, the spine is one of the most common places to get hurt.^[1] Both mistakes are avoidable.

What the research says

A few terms first. A systematic review collects every study on a question in a set way. A meta-analysis pools their results into one estimate. EMG (electromyography) measures electrical activity in a muscle during a lift. It shows which muscles worked harder in the moment, not which ones grow more over months.

Competitive lifters get hurt less often than contact-sport athletes, but the back is a common site

Limited evidence on the exact numbers.

A rounded back hasn't been shown to cause back pain

"Never round your back" is the most repeated deadlift rule. The research hasn't backed the fear behind it.

A 2020 meta-analysis gathered 11 studies of how much people bent their lower back while lifting. Nine of the 11 found no clear difference between people with back pain and people without.^[2] On one measure, the people with back pain bent their lower back about 6° less, not more.^[2] The authors called it "low-quality evidence" that more bending while lifting was not a risk factor for back pain.^[2] Limited evidence.

That's a long way from proof that rounding is safe. The review was about lifting in general, not heavy deadlifts.^[2] Only one study followed people over time.^[2] And a study done at one moment can't tell whether the way people bent came before the pain or after it.

A wider review read 41 reviews on posture, bending, twisting and heavy physical work. Some reported a link with back pain and some didn't. The authors found no agreement that any of them causes back pain, and they point out that a link isn't a cause.^[3] Mixed evidence. The studies disagree. They looked at posture and physical work in general, not at deadlifts.^[3]

So: a rounded back hasn't been shown to cause back pain. Nobody has tested heavy rounded-back deadlifts and shown them to be safe either.

Deadlifts have been used to treat back pain

This part is about people who already had back pain. It shows that, in one closely supervised trial, people with back pain could learn the deadlift. It isn't advice to treat your own back.

A Cochrane review is a large, careful type of systematic review. This one included 249 trials of exercise for adults whose lower back pain had lasted more than 12 weeks.^[4] Exercise probably eased pain more than no treatment or usual care. The gain in day-to-day function was smaller.^[4] Moderate evidence.

In one Swedish trial, 70 adults with recurring lower back pain were split into two groups.^[5] One group learned the deadlift from a physiotherapist, in small groups of 2 to 6 people.^[6] They started at 10 kg, bar included, with the back held neutral. The weight went up only once the technique was right.^[6] The other group did gentle, low-load exercises for control of the trunk. Both groups were offered 12 sessions over 8 weeks.^[5] The deadlift group came to about 11 on average, and the gentle-exercise group to about 6.^[6]

Pain fell in both groups, and by about the same amount. The deadlift group did not do better. On one measure of daily activities, it did worse than the gentle-exercise group.^[5] Two years later there was still no difference between the groups. In both, between half and four in five people reported less pain and disability.^[6] Two people in the deadlift group reported side effects, and one of them dropped out.^[6]

Limited evidence. It's one trial and its follow-up. The participants were 22 to 60 years old, 31 men and 39 women, and every session was supervised.^[6]

Sumo and conventional are both legitimate

In one study of national-level powerlifters, 12 lifted sumo and 12 lifted conventional. The sumo lifters stood more than twice as wide (about 70 cm against 32 cm) and turned their feet out more (about 42° against 14°). Their torso and thighs were 5–10° more upright. The conventional lifters moved the bar about 25–40% farther and did about 25–40% more work per rep. The study found no clear difference in the load on the hips.^[7] Limited evidence. It's one small study, with different lifters in each style.

That shorter bar path may be why some lifters call sumo "cheating". It's a different technique, and lifters in that study used both styles at a national championship.^[7]

What about the lower back? A 1991 study of 57 national-class powerlifters (13 women, 44 men) estimated the load on a joint low in the spine with a computer model. Sumo lifts put about 10% less bending load on that joint than conventional lifts.^[8] Limited evidence. It's one study, with modelled numbers rather than measured ones, comparing different lifters. The authors also found large differences between lifters in how they loaded their joints.^[8] And nobody has studied injury rates in the two styles well yet. So "sumo is safer" isn't supported, and neither is the opposite.

The two styles also work the muscles a little differently. One study tested 13 college football players at their 12RM, the most weight they could lift 12 times. The sumo deadlift showed more EMG activity in two of the quadriceps muscles (the front of the thigh) and in a shin muscle. The conventional deadlift showed more in the calf. The authors suggest picking the style by which muscles matter most to you.^[9] Limited evidence. Remember, EMG shows which muscles worked harder during the lift. It doesn't prove more growth over months.

The trap bar shifts the load toward the knees

A lab study tested 19 male powerlifters. The trap bar put less peak load on the lower back, hips and ankles than a straight bar, and more on the knees. Their 1RM was higher too. That's your one-rep max: the most weight you can lift once. It averaged about 265 kg on the trap bar and 245 kg on the straight bar.^[10] Limited evidence. It's one study in experienced men. Less load on the lower back hasn't been tested as "fewer injuries".

The Romanian deadlift leans on the hamstrings

A systematic review of 19 EMG studies looked at the deadlift and its variants. The hamstrings are the muscles at the back of the thigh. In trained adults, the lower-back muscles and the quadriceps were usually more active than the glutes and the hamstrings. The Romanian deadlift was the exception: there, the hamstrings were more active than the lower-back muscles.^[11] Limited evidence. Most of the studies used men.^[11] The studies measured EMG in different ways, some disagreed, and they couldn't be pooled.^[11]

Does that activity turn into growth? One trial suggests it can. It put 36 people who didn't do resistance training on a 9-week program, three times a week. One group did the stiff-leg deadlift and one did nothing. A third did the Nordic curl: you kneel, someone holds your ankles, and you lower your body forward slowly. The stiff-leg deadlift grew the whole hamstrings by about 7%, and the Nordic curl by about 11%. The authors called the growth similar. Each exercise grew a different part of the muscle.^[12] Limited evidence. It's one small trial, in beginners, over 9 weeks.

How to apply it

Pick your first deadlift

The research doesn't pick a winner, so pick the version you can control. This is coaching practice, not a tested finding.

• Start with a hinge you can control. A Romanian deadlift with dumbbells or a light bar, or a trap-bar deadlift, are good places to begin.
• Learn the pull from the floor if you want it. Conventional or sumo, both are fine.
• Choose sumo or conventional by feel. Pick the one that feels more natural for your build. Nobody has studied matching style to body shape well yet.

Watch it first

Watch a demonstration before your first set. Watch it twice: once for the bar, once for the hips. Exercises in the Gymbers app come with a video, and you can save the videos to your phone.

The technique checklist

These cues are coaching practice. Use one at a time.

1. Keep the bar close to your legs on the way up and on the way down.
2. Push your hips back to start the hinge, rather than bending your knees first.
3. Brace your trunk (tighten the muscles around your middle) before the bar leaves the floor.
4. Push the floor away as you stand up.
5. Stand tall at the top. Don't lean back.
6. Lower the bar under control.

For sumo, set your feet wide with your toes turned out, like the lifters in the study above. Coaching practice.

In the session

These steps are coaching practice too.

1. Start light. An empty bar, dumbbells, a kettlebell or a trap bar. In the back-pain trial, people started at 10 kg, bar included.^[6] That shows a light start was used in a trial, not that it prevents injury.
2. Ramp up first. Do a few lighter sets before your working sets (see how to warm up for lifting).
3. Leave 2–3 reps in reserve. RIR (reps in reserve) is how many more reps you could have done when you stopped. While you learn the lift, end each set with 2–3 left (more in how hard a set should be).
4. End the set when a rep looks different. Hips shoot up first, the back rounds more than on rep one, or the bar drifts forward: that was the last rep.
5. Add weight once every rep looks the same. Then add a little and check again. Learning the lifts covers filming a set to check yourself.
6. Pair the hinge with a leg curl. For the back of the thigh, a Romanian deadlift and a seated leg curl make a fair pair (see how to choose exercises).

A worked example

Here's a 70 kg beginner learning the Romanian deadlift. Three sets of 8, after two ramp-up sets.

• Session 1: empty bar (20 kg) × 8, 8, 8. Every rep looks the same.
• Session 2: 30 kg × 8, 8, 8. Still clean.
• Session 3: 40 kg × 8, 8, 7. On the eighth rep of set 3, the bar drifts away from the legs. He stops there.
• Session 4: 40 kg × 8, 8, 8. Clean. Next session, 45 kg.

The numbers are an example, not a schedule.

What's practical versus optimal. These first sets will feel easy. That's the trade: a few easy sessions for a hinge you can repeat under load. Coaching practice.

Common mistakes

Pulling heavy from the floor in week one. Someone at the next platform may pull a big number, and it's tempting to match it. Start with a hinge you can control and add weight once the reps match.

Treating sumo as cheating, or as the safe choice. Both ideas point to real differences: a shorter bar path and a little less modelled load on the lower back.^[7][8] Neither makes it cheating or safer. Pick the style that feels better and lift it well.

Forcing out reps that look different. The last rep feels like the one that counts. But when the hips shoot up or the bar drifts forward, it's a different lift. End the set there and log what you did.

Leaning back at the top. It feels like a stronger finish. Stand tall and stop. Coaching practice.

Using deadlifts to fix your own back pain. The trial in people with back pain had a physiotherapist lead every session, and the deadlift did no better than gentle exercise.^[5][6] If your back hurts, see a doctor or physiotherapist first.

How Gymbers fits

A deadlift is learned in many light sets. Exercises in the Gymbers app come with a video, and the app can add warm-up (ramp-up) sets before your working sets. The log records the weight and reps for every set, and last session's numbers sit beside each set. You can log RIR next to each set, so you can see how many reps you left each time.