Some components of generalized anxiety disorder, such as physical symptoms, are thought to reflect autonomic nervous system arousal. This study primarily assessed the relationships between psychophysiological and clinical measures using venlafaxine extended release or applied relaxation, and secondarily, the impact of combination treatment in patients not remitting after 8 weeks. Fifty-eight patients were randomly assigned to 8 weeks of treatment with either venlafaxine or applied relaxation (Phase I). Non-remitted patients received combination treatment for an additional 8 weeks (Phase II). Assessments included the Hamilton Anxiety Scale (HAM-A), Beck Depression Inventory, Penn State Worry Questionnaire and the Stroop Color-Word Task coupled with electrophysiological measures (skin conductance and frontalis electromyography (EMG)). In Phase 1, a time effect was found for the clinical and skin conductance measures. Thirteen patients from each group were in remission. In Phase 2, seven additional patients remitted. Baseline psychophysiological measures were not associated with baseline clinical variables or with clinical outcomes. Independently of treatment allocation, a reduction in frontal EMG values at week 4 was significantly associated with a decrease in HAM-A scores at week 8. At week 4, responders from the applied relaxation group had lower electrophysiological activity than the venlafaxine group. Baseline psychophysiological measures were not linked with clinical measures at study inclusion or with treatment response. Frontal EMG response at week 4 is a possible predictor of treatment response. Treatment combination enhances treatment response after initial failure.