The purpose of this investigation was to test the hypothesis that subjects (Ss) can exert voluntary, bidirectional heart rate (HR) control in the absence of concomitant changes in respiration rate (RR) and muscle activity (chin EMG). Feedback-assisted HR control was studied under four experimental conditions involving increasing levels of somatic restraint: (a) No somatic controls, (b) RR pacing, (c) EMG feedback, to maintain constant muscle activity, and (d) RR pacing plus EMG feedback. Fifteen Ss, eight females and seven males, were assigned to each of these four conditions (N = 60). The HR control task involved ten increase and ten decrease trials in random order, each lasting 60 sec and preceded by a 60-sec resting baseline period. The results indicated that Ss produced significant HR increases (Xdelta = +3.4 bpm), but were unable to effect decreases (Xdelta = -.3 bpm). Furthermore, the magnitude of the obtained HR changes did not differ by experimental condition, i.e., by level of somatic restraint. While concomitant bidirectional RR changes were found, the pacing procedure effectively reduced, and on HR increase trials, nearly eliminated, RR involvement. No increases in EMG activity were found on HR increase trials for any of the four experimental conditions, whereas all groups showed some EMG decreases on HR deceleration trials. With respect to the question of somatic mediation of autonomic control, it is concluded that HR increases produced under paced respiration "favor" an interpretation of unmediated HR acceleration.