Unstable Surface Training Is Only for the Injured or Deconditioned

A new study suggests if you’re going to use stability balls for the bench press, you should ditch the barbell in favor of dumbbells.

For the average trainee, unstable surface training might not be the best idea. However, it does have a place in rehab and training beginners. A recent study in the Journal of Strength and Conditioning Research investigated how to get the most out of unstable surface training.

Unstable surface training increases the demand on stabilizing muscles. As such, the researchers noted that while it may reduce the load too much during barbell lifts, combining unstable surface training with uncoupled resistance like dumbbells may actually increase the demand on the core enough to make it worthwhile.

Study Design

In the study, the researchers used EMG data to record muscle activation during three different tests. All three tests were performed on a Swiss ball. Each of the following tests was performed for five reps:

  • A barbell bench press with fifty percent of one-rep-max (1RM)
  • A dumbbell bench press with the same load
  • A dumbbell bench press with 25% of 1RM

EMG data was gathered for four muscles: the pectoralis major (pecs), the anterior deltoid (delts), the triceps brachii (triceps), and the rectus abdominus (abs). The first three act as prime movers during the bench press, meaning they engage to move the weight. The abs work isometrically during this exercise to stabilize the body. Because of the unstable surface, the abs must work harder to provide stability.

The loads used were percentages of a traditional 1RM flat bench. Bear in mind, although the loads look low, they aren’t quite as low as they seem. Dumbbell bench press numbers would be less than a traditional barbell press even without the Swiss ball, and these numbers would drop further on an unstable surface. So, while fifty percent of 1RM was a pretty low bench press value to be working with, it was intensified by the use of dumbbells and Swiss balls.


For the shoulders and triceps, the dumbbell and barbell lifts of the same weight were about equal in muscle activity. The dumbbells performed better for the pecs and the abs. This is because the uncoupled nature of dumbbells created a need for the dumbbells to not fall laterally, which engaged the pecs more. The dumbbells also decreased the stability of the movement and required greater abdominal activation.

In all the lifts, however, the triceps were by far the most activated of the muscles, experiencing a roughly seventy-percent greater response than the next most activated muscle, the delts. The 25% dumbbell condition resulted in greatly reduced activity in all the muscles studied.


Individuals who are rehabbing an injury, in a deconditioned state, or in a completely unconditioned state will probably benefit most from unstable surface and uncoupled resistance training. The researchers noted that in untrained subjects, unstable surface training resulted in about the same strength gains as stable surface training. This reinforces the idea that beginner’s gains come regardless of the activity. There might be added benefit from the increased demand on the core, however.

Another important note is that the load used affects the instability of an unstable surface like a Swiss ball. In other words, a heavier load will squish the ball more, and in effect may increase its degree of stability. This may explain why the jump in abdominal activity seen from the light dumbbells to the heavier ones was lower than the jump in shoulder and pec activity. A heavier load than the one used in this study might have diminishing returns.

The major take-home point was that load had the biggest effect on activation. As such, reducing the maximum load by introducing instability is a bad idea for anyone who is not untrained, deconditioned, or rehabbing an injury.


1. Brian Campbell, et. al., “An Evaluation of Upper-Body Muscle Activation During Coupled and Uncoupled Instability Resistance Training,” Journal of Strength and Conditioning Research, 28(7), 2

Photo courtesy of Shutterstock.