This paper reports the relationships among changes in cardiovagal activity, surface EMG, and measures of pulmonary function in a study of relaxation therapy for asthma. Changes in FEV1/FVC were negatively correlated with those in cardiac interbeat interval, consistent with the hypothesis that relaxation-induced changes in airway function are mediated autonomically, with increased vagal tone and/or decreased sympathetic arousal producing bronchoconstriction. Contrary to Kotses's theory of a vagal-trigeminal reflex as mediator for relaxation-induced improvement in asthma, decreases in pulmonary function occurred during relaxation sessions, accompanied by increases in cardiovagal activity, and within-session changes in frontal EMG in the first session of training were positively associated with changes in a measure of pulmonary function (FEV1/FVC). However, consistent with this hypothesis, first-session frontalis EMG changes were positively associated with changes in respiratory sinus arrhythmia, and last-session changes in cardiac interbeat interval were positively associated with changes in FEV1/FVC. The results suggest that the immediate effects of generalized relaxation instruction can be associated with a parasympathetic rebound, which, in tum, may induce countertherapeutic changes in asthma. However, the effects of specific facial muscle relaxation remain unclear.