ABSTRACTThree groups of 10 normal subjects were trained with 13 maximal isometric contractions on each of 8 days. During 8 of the 13 trials, subject groups received either myoelectric feedback (MFB), or force feedback (FFB), or no feedback. Over the 8 training days all groups increased EMG activity, but no significant group differences in rate of acquisition were obtained. All groups made gains in force over the 8 days. Force feedback produced better acquisition than either MFB or control conditions. However, both MFB and FFB produced comparable and significant differences between feedback and no‐feedback trials within a session, highlighting the importance of control group designs in the assessment of biofeedback effects. Dissociation between surface EMG and force was shown not only in the training effects, but also in the difference between feedback and no‐feedback performance within a session. An analysis of covariance revealed that during feedback trials MFB produced some EMG facilitation which could not be attributed to increased force production. This effect may have been due to selective reinforcement by MFB of co‐contraction in the extraneous musculature, or of changes in motor unit firing patterns. These phenomena may also account for the relative failure of MFB to facilitate force acquisition. A one‐month follow‐up showed significant retention following similar losses of force in a