Upper Back Workouts: Complete Exercise Guide

Upper back workouts in three numbers

Three measured figures set the rules for upper back workouts. One is on grip, one on volume, one on the pull-up grip that loads the mid traps hardest.

Study figures come from the 16 PubMed papers in the sources list. Activation data are acute electromyography, which shows which muscle works in a movement, not how much it will grow. Workouts are templates.

What this upper back workout guide covers

Upper back muscles: what you are actually training

The upper back muscles are five groups with different jobs, and upper back workouts have to cover all of them. The trapezius has upper, middle and lower fibres that elevate, retract and depress the shoulder blade. The rhomboids sit under the middle traps and retract. The rear deltoid extends the arm. The latissimus dorsi pulls the arm down and back. The teres major assists it.

Thickness from the side comes from the middle trapezius, rhomboids and rear delts, which the horizontal rows in upper back workouts load. Width from behind comes from the lats and teres major, which vertical pulls load.

The electromyography on upper back exercises confirms the split. The seated row produced the highest middle trapezius and rhomboid activity of four lat exercises. The wide grip pulldown gave the best lat to biceps ratio (PMID 15228624).

The scapular muscles also decide how the shoulder moves. Altered scapular motion, with reduced serratus anterior and raised upper trapezius activity, appears across impingement, rotator cuff tendinopathy and instability (PMID 19194022).

That is why upper back workouts are posture and shoulder health work as well as mass work. The lower and middle trapezius are the muscles that balance the upper trapezius in those patients (PMID 19194022).

Best upper back exercises: six with activation data

These six are the best upper back exercises by one test: each has at least one electromyography study behind it. The ranking of upper back exercises is by what was measured, not by tradition.

1. Inverted row. Of three rows compared, the inverted row produced the highest activation of the lats, upper back and hip extensors. It also had the lowest lumbar spine load (PMID 19620925). In 26 healthy adults it also recruited the lower trapezius and serratus anterior, the muscles that balance the upper trap (PMID 29944029).

2. Seated cable row. The highest middle trapezius and rhomboid activity of the four exercises tested, and a high lat to biceps ratio. Actively squeezing the shoulder blades together did not raise mid trap activity (PMID 15228624).

3. Bent over barbell row. Large, symmetrical activation across the whole back, and also the largest lumbar spine load of the three rows (PMID 19620925).

4. Pull-up, pronated grip. Middle trapezius peak activation was 60.1 percent of maximum with a pronated grip against 37.1 percent with a neutral grip. Lat activation was similar across grips (PMID 28011412).

5. Lat pulldown, bar in front. An anterior bar path was the configuration most consistently linked to higher lat activation across 23 studies. A pronated grip beat a supinated grip (PMID 42647355, 20543740).

6. Prone barbell row. Rear deltoid and middle trapezius activation did not differ by range of motion; lats worked harder in the upper half (PMID 40513198).

Rhomboid exercises are the same list. The rhomboids sit under the middle trapezius and share its electromyography signal. The seated row and pronated pull-up that load the mid traps load the rhomboids too (PMID 15228624).

Grip width, hand position and range: what changes and what does not

Wide grip for lats is the oldest rule in upper back workouts, and the direct test did not support it. Fifteen men pulled a six repetition maximum with narrow, medium and wide pronated grips, 1, 1.5 and 2 times shoulder width (PMID 24662157).

Narrow and medium grips were stronger, 80.3 and 80.0 kg against 77.3 kg wide, which matters for loading upper back workouts. Lat, trapezius and infraspinatus activation did not differ across widths over the whole movement (PMID 24662157).

Hand orientation mattered more than width in these upper back workouts. In 12 men, a pronated grip produced higher lat activity than a supinated grip on the pulldown. The middle trapezius and biceps did not change (PMID 20543740).

Bar path mattered most. Across 23 studies, pulling the bar to the front was the configuration most consistently linked to higher lat activation. Grip width results were inconsistent (PMID 42647355).

On the pull-up, grip changed the traps more than the lats. Pronated grip raised mid trap peak to 60.1 percent of maximum against 37.1 percent neutral; lat activation was similar across grips (PMID 28011412).

Range of motion on the prone row, a staple of back thickness exercises, changed little. Lats worked harder in the upper half of the movement and the middle trapezius peaked lower there. Rear delt and biceps did not differ (PMID 40513198).

Squeezing the shoulder blades on a seated row is the cue in every gym. It did not change middle trapezius or rhomboid activity in the one study that tested it (PMID 15228624).

Three rows compared: activation and spine load

A spine biomechanics laboratory compared the inverted row, the standing bent over barbell row and the standing one arm cable row in seven men. It measured back and hip activation, spine position and joint moments (PMID 19620925).

The inverted row won on activation and lost nothing on the spine, which makes it the best value of the upper back exercises for volume. It produced the highest lat, upper back and hip extensor activity, and the lowest lumbar erector activity, matching the lowest spine load measured (PMID 19620925).

The bent over row produced large, symmetrical activation across the back and the largest lumbar spine load of the three (PMID 19620925). That is the trade in most upper back workouts.

The one arm cable row challenged the trunk in rotation more than it loaded the back. That makes it a core exercise with a row attached (PMID 19620925).

The spine load matters in upper back workouts because the erectors are already working. In a squat at 80 percent of maximum, lower lumbar erector activity exceeded the deadlift by 34.5 percent. The deadlift loaded the upper lumbar erectors 12.9 percent more than the squat (PMID 18076231).

Exercises that load the lumbar extensors all activate them, but chronic adaptation depends on pelvic restraint and specificity. A heavy bent over row is not a substitute for direct lower back work (PMID 24092889).

Source for the table: PMID 19620925.

Two complete upper back workouts

Both upper back workouts follow the same pattern. One heavy horizontal row, one vertical pull, one lighter row for volume, one rear delt or lower trap movement. The first is the gym version with barbells and cables.

The second upper back workout swaps the barbell for a pulldown and a chest supported row. It is for a day when the lower back needs a rest or the session follows deadlifts.

Each of the two upper back workouts lands between 17 and 20 sets. Run one of the upper back workouts twice a week, or both once each. The weekly total then sits in the range the volume meta-analysis rewarded (PMID 27433992).

Upper back exercises with dumbbells only

Upper back exercises with dumbbells cover every job on the muscle table with a pair of dumbbells and a bench. The movements below are the dumbbell versions of the best upper back exercises above, chosen for the same activation pattern.

The pullover is the weak link of dumbbell upper back workouts. Across the review, it preferentially activated the pectoralis major rather than selectively targeting the lats. It is a pec and lat exercise, not a pulldown substitute (PMID 42647355).

A door frame, a table edge or a suspension strap fixes that gap. The inverted row needs no dumbbells and produced the highest lat activation of the three rows tested (PMID 19620925).

How many sets and reps for the upper back

Weekly volume for upper back workouts has a measured dose response. Across 15 studies and 34 groups, each additional weekly set added 0.37 percent to the gain in muscle size. Higher volume groups gained 3.9 percent more than lower volume groups (PMID 27433992).

Ten or more sets a week per muscle showed a trend over fewer (PMID 27433992). The 2026 meta-regression of 67 studies and 2058 people found growth and strength both rose with volume, with diminishing returns (PMID 41343037).

Frequency of upper back workouts matters less than the total. Across 25 studies, training a muscle once, twice or three times a week produced the same growth once weekly sets were equal (PMID 30558493).

Strength is the exception for upper back workouts. In the 2026 analysis, strength gains rose with frequency, with diminishing returns, while the frequency effect on size was compatible with nothing (PMID 41343037).

Proximity to failure in upper back workouts cuts the other way. Across the exploratory meta-regression, muscle size increased as sets ended closer to failure. Strength was similar across a wide range of reps in reserve (PMID 38970765).

For upper back workouts that means 10 to 20 hard sets a week, split as you like. The rep range in the workouts above runs 6 to 20 because the row loads heavy and the rear delt work does not.

Upper back training and posture

Rounded shoulders and a forward head are the reason many people search for upper back workouts and rhomboid exercises. The shoulder literature gives that search a mechanism.

Scapular kinematic changes are found in impingement, rotator cuff tendinopathy, instability and frozen shoulder. The pattern is reduced serratus anterior and increased upper trapezius activation (PMID 19194022).

The same changes appear in healthy people with a short pectoralis minor, tight posterior shoulder tissue, or excessive thoracic kyphosis. That last one is the desk posture (PMID 19194022).

The inverted row, in both upper back workouts above, recruits the muscles on the other side of that imbalance. Upper trapezius to lower trapezius ratios ran 1.6 to 2.2 and upper to serratus 1 to 4.7 in 26 healthy adults. The spinal stabilisers were active throughout (PMID 29944029).

Core stability frames the whole thing. The trunk controls position and motion over the pelvis so force transfers to the arms. Upper back strength is part of that control (PMID 16526831).

No trial on this page shows that upper back workouts or rhomboid exercises fix posture. The mechanism is measured; the outcome is an inference, and this page says so.

Five upper back workout mistakes the studies rule out

Verdict: upper back workouts in 16 studies

1. Rows are the back thickness exercises, pulls build width. Seated row for mid traps, pulldown for lats (PMID 15228624).

2. Wide grip is not a lat grip. Same activation, 3 kg weaker (PMID 24662157).

3. Pronated beats supinated for lats. On the pulldown and on pull-up mid traps (PMID 20543740, 28011412).

4. Bar to the front. The most consistent lat configuration across 23 studies (PMID 42647355).

5. The inverted row is the best value row. Highest activation, lowest spine load (PMID 19620925).

6. The squeeze cue adds nothing measurable. Mid trap activity unchanged with active retraction (PMID 15228624).

7. Each weekly set adds 0.37 percent. Higher volume gained 3.9 percent more (PMID 27433992).

8. Frequency is preference. Same growth at equal volume across 25 studies (PMID 30558493).

9. Closer to failure grows more. Strength did not care (PMID 38970765).

10. Upper back work balances the shoulder. Lower trap and serratus against the upper trap (PMID 19194022, 29944029).