Three groups of 10 subjects were exposed to intermittent bursts of loud noise. Two groups attempted bi-directional cardiac control immediately prior to the noise bursts, one with the aid of continuous heart-rate biofeedback and one without. The third group did not attempt heart-rate control but instead received false feedback that heart rate was either increasing or decreasing prior to the noise bursts. Only the biofeedback-assisted group demonstrated reliably different cardiac changes during attempted heart-rate increase and decrease. The heart rates of the false-feedback subjects were unaffected by the direction of change signalled by the biofeedback display. However, a significant differential effect on subjects' ratings of the noise bursts, in terms of loudness and unpleasantness, was apparent only in the false-feedback group. Skin conductance responses showed no such effect in any of the three groups. The results are most readily accounted for in terms of attribution theory.