ABSTRACTTwenty‐two normal adults participated in an electromyographic (EMG) biofeedback experiment designed to test if the feedback stimulus is necessary in obtaining EMG reductions during typical laboratory procedures, and if prior training on one muscle facilitates the training of a second muscle. One group of subjects received forearm feedback training followed by frontalis training. A second group received training in the reverse order. Two control groups relaxed first on their own followed by either forearm or frontalis training. Attention was directed toward motivating control subjects to perform maximally during relaxation without feedback. Heart and respiration rates and skin conductance and temperature were also recorded. Both trained and untrained subjects produced significant EMG reductions but did not differ from each other; nor did a transfer of training effect emerge. No differences resulted between feedback and non‐feedback conditions for the other physiological measures or for changes in state anxiety. These data compromise somewhat the previous demonstrations of EMG biofeedback “learning,” and are unsupportive of EMG biofeedback as a general relaxation training technique.