The best biceps exercises: why the measurement changes the answer

Three variables hide inside one curl

“Incline curls stretch the biceps” is a plausible starting explanation. It is not a complete description of the exercise. Shoulder position changes one part of the task; elbow angle changes another; the direction and magnitude of the external resistance change a third. Comparing curl names without those details can mistake a whole configuration for a single variable.

The biceps brachii crosses the shoulder and elbow. Other elbow flexors also contribute to moving the forearm. That is why “arm growth,” “elbow-flexor thickness” and “biceps volume” should remain separate labels. A measurement taken over several tissues cannot tell you precisely which individual muscle grew unless the method distinguishes them.

For a useful curl comparison, ask three questions. Where is the upper arm relative to the torso? Through which elbow angles does it move? Where does the exercise impose the greatest external torque? A dumbbell and a cable can place the hardest part at different points even when the arm position looks similar.

These questions do not require an elaborate program. Their value is to help you choose a small number of repeatable movements and understand what changed when you swap one. You can start with one accessible curl, learn it well, and add a second configuration when you have a clear reason.

Preacher versus incline: the regional result

Kassiano and colleagues compared preacher and incline dumbbell curls in sixty-three young women over eight weeks. The reported outcome was elbow-flexor thickness at regional measurement sites, together with exercise-specific strength. The accessible abstract provides a useful contrast: incline favoured the proximal site, while preacher favoured the distal site. [01]

We show the between-condition difference on the same axis. Positive values mean a greater thickness change with incline; negative values mean a greater change with preacher. This is not a graph of total biceps volume, and it is not a percentage of fibres activated. The full paper was inaccessible, so our description stays within the public abstract rather than inventing additional methods.

The middle-site estimate is not plotted because the accessible report did not provide it. Leaving a gap is better than inserting a zero that looks measured. A zero would imply a precise estimate of no difference, which is a claim we do not have evidence to make.

The practical lesson is regional variation. An exercise can compare favourably at one site without winning everywhere. It also means “best curl” depends on whether you are asking about a particular region, a whole muscle, total elbow-flexor size or performance on the curl itself.

New MRI research makes the muscle distinction clearer

A study published in October 2026 used MRI to distinguish individual elbow-flexor muscles after incline and preacher dumbbell training. Twenty-one untrained adults trained opposite arms with different exercises for twelve weeks. The public abstract reported different patterns of change in biceps, brachialis and brachioradialis, while the total flexor-volume comparison was not clearly separated. [02]

That is valuable because it addresses a measurement problem. A similar total change can conceal different contributions from the muscles underneath it. The study is recent, small and available to us as an abstract; it should inform the conversation without becoming the final authority on every curl configuration.

These are reported condition changes, not guaranteed outcomes for a reader. The abstract supplies no confidence intervals for the percentages. The total comparison had p = .134, which does not prove exact equivalence. Exercise configurations also changed more than shoulder position, so the results cannot be attributed exclusively to a stretch explanation.

This is a good example of how a more precise measurement can improve a recommendation without creating a simplistic winner. If the goal is balanced elbow-flexor development, more than one configuration can make practical sense. If the goal is a particular muscle, evidence that actually measures that muscle deserves extra attention.

Shoulder extension is not an automatic upgrade

The behind-the-body cable curl is often recommended because the upper arm sits behind the torso. That position can change biceps length, but the cable also creates a particular resistance profile. To investigate the position itself, studies need to control other important aspects of the task rather than comparing every difference at once.

Attarieh and colleagues compared shoulder-flexed and shoulder-extended cable-curl conditions with matched resistance profiles. Fifteen men completed ten weeks of training. The study separately sampled biceps and brachialis and did not detect a clear between-condition difference in the reported growth outcomes. Small samples leave uncertainty; no detected difference is not proof that all cable setups are identical. [03]

Larsen and colleagues' 2026 trial compared neutral and extended shoulder positions during cable curls with a matched descending resistance profile. Twenty-four men completed the ten-week protocol. Its combined elbow-flexor measurements did not show a meaningful advantage for extension under the tested setup. Again, those measurements should not be relabelled as isolated biceps volume. [04]

Taken together, these papers make a universal claim harder to defend. Shoulder extension can be a reasonable option. It is not a setting that automatically upgrades every curl. Comfort, resistance profile, available range and your ability to repeat the task still matter.

Long lengths and partial repetitions need precise language

“Train at long muscle lengths” can refer to several different interventions. You might change shoulder position while keeping the elbow range similar. You might change which elbow angles receive training. You might keep a full range but shift where external torque peaks. Those experiments do not all test the same thing.

Pedrosa and colleagues compared initial and final portions of a preacher curl in nineteen untrained women. The initial-range condition favoured a distal biceps cross-sectional-area outcome, while other measured size comparisons did not clearly separate. The experiment did not include a full-range training group. It therefore cannot establish that lengthened partials beat full-range curls. [05]

Nunes and colleagues used cable and barbell preacher curls to change the resistance profile while keeping full-range training. They did not detect a clear difference in the measured midpoint thickness outcome. That is a different study from a partial-range comparison, despite the similar language about torque and muscle length. [06]

For a starting program, use a comfortable range you can repeat and progress. More specialized partial-range work can be considered as a distinct programming choice, but it should not be sold as mandatory on the basis of a trial that answered a different question. Keep the intervention and outcome aligned when reading a headline.

A small study should stay a small study

Older regional curl research also illustrates why careful reading matters. Zabaleta-Korta and colleagues studied incline and preacher training in women with some training experience. One regional within-group change was significant, but the reported between-group comparisons did not clearly establish a winner. The paper also contains reporting inconsistencies that make a precision growth chart inappropriate. [07]

The difference between “this group improved” and “this group improved more” is not a technicality. A group can pass a significance threshold while another does not, without a demonstrated difference between the groups. If an article converts that situation into a confident exercise ranking, the conclusion has become stronger than the experiment.

Look for the actual comparison, sample, endpoint and uncertainty. Was the paper studying novices? Did it measure one site or an entire muscle? Were the arms trained within the same person? Did both exercises receive similar effort and sets? Did other pulling exercises occur in the program? Those details explain what a result can support.

You do not need to become a statistician to choose a curl. You do need to be cautious about a claim that turns one small trial into an exact blueprint for every arm. The study can still offer a useful clue while leaving the final programming choice open.

The mind-muscle connection is an optional cue

Focusing attention on the target muscle can help some lifters execute a movement consistently. It can also become another task to perform perfectly, distracting from a repeatable range and progression. Use the cue if it helps the movement, rather than judging every set by whether it produced an intense sensation.

Schoenfeld and colleagues compared internal and external attentional-focus strategies over eight weeks in untrained men. The accessible abstract reported a larger elbow-flexor thickness change with internal focus, without the same advantage in the quadriceps. The study does not establish that an internal cue is required for all muscles or every trained lifter. [08]

An acute change in signal also does not provide a growth guarantee. The EMG interpretation problem applies here just as it does to exercise rankings. A cue may change how you perform a task, while your decision still needs to consider whether execution, loading and repeatability improve. [09]

Try a simple cue for several sets and observe the practical result. If it helps you avoid swinging and keep the intended elbow movement, retain it. If it makes you slow the exercise excessively or forget the progression rule, choose a different cue. The useful outcome is better training, not a stronger belief in a sensation.

Choose curls by the task they provide

This shortlist describes roles rather than scientific rankings. A cable label alone does not specify its resistance profile; moving your body relative to the pulley changes the task. A preacher setup can also differ between implements and machines. Record the version you actually use.

Choose one main curl first. If you add another, make the reason explicit: a different configuration, easier progression, greater comfort or a distinct development priority. Do not add four exercises simply to cover four fashionable names while leaving the dose and endpoint undefined.

An illustrative two-day arm plan

The example below provides six direct curl sets across two exposures. This is an original starting template, not a demonstrated optimum. Your pulling program already involves elbow flexors to some degree, so the appropriate amount of direct work depends on the rest of the week. Fractional accounting in research is a model for indirect contribution, not a device that measures the exact stimulus of each row. [10]

Use the first sessions to establish a reproducible setup. Keep the upper-arm position, range and forearm orientation consistent. Stop the set at an endpoint you can recognize. Training to failure is an option, not a requirement that overrides execution or the rest of the program. [11]

If your back sessions already leave substantial elbow-flexor fatigue, begin more conservatively. If direct curl performance improves and recovery is manageable, retain the dose rather than adding sets automatically. If progress stalls across comparable sessions, examine the load increments, pulling volume and schedule before deciding that a completely different curl is necessary.

You can also use one configuration on both days. The evidence does not establish a compulsory preacher-plus-incline pairing. The two-exercise example gives you different tasks to evaluate, while simplicity can be a legitimate advantage for a beginner or a limited-equipment setting.

Make progression honest enough to learn from

Choose a repetition container such as eight to twelve for the main curl. Add controlled repetitions at the same load while maintaining the planned endpoint. When you consistently reach the upper range, make a modest load increase and build the repetitions again. Keep the task stable enough that the log reflects an actual improvement.

If every load increase requires a larger torso swing or a shorter elbow movement, record that honestly. It may be a different training choice, but it is not a clean record on the old one. A cable stack from another machine or a new preacher angle also deserves a fresh baseline rather than a celebratory comparison with the previous setup.

Evaluate several weeks, not one pump. Watch repeatable performance, the amount of direct and indirect work, and whether the next session is recoverable. If you measure your arm, use consistent conditions and remember that circumference includes more than the biceps. It is not interchangeable with the MRI or ultrasound outcomes in the papers.

The current evidence supports a flexible menu and more precise questions. It does not require abandoning a productive basic curl, chasing a single shoulder position or pretending one regional result describes the entire arm. Choose a movement you can repeat, understand what it changes, and let a well-kept training record guide the next adjustment.