The etiological tinnitus models propose that suffering can be caused and aggravated by heightened physiological arousal. Therefore psychophysiological treatments are applied. Stability of the measured parameters is essential for the use of biofeedback as well as to permit the attribution of changes to the administered treatment. The aim of our study was to investigate the 3-month reproducibility of psychophysiological parameters in 60 tinnitus patients. Using a repeated-measures design, the activity of these parameters was assessed twice during various stress and relaxation trials. The results showed that the measurements of frontalis, masseter and trapezius muscles were stable, while for the sternocleidomastoid, the skin conductance level (SCL) and the skin temperature retest-stability could not be evidenced. For all parameters, test–retest stability was weak for the relative scores. In conclusion, our study has important implications for applied psychophysiology research: (1) the measurement of EMG assessed in a clinical sample is stable over a 3-month interval; (2) in contrast, the measurements of SCL and skin temperature as well as all relative scores are less stable; and (3) the stability of EMG-parameters in our sample gives first hints that physiological changes can be attributed to an administered biofeedback treatment but further research is required.