In this study, three types of treatments for chronic musculoskeletal pain were compared. Fifty-seven patients who suffered from chronic back pain and 21 patients who suffered from temporomandibular pain and dysfunction were randomly assigned to either electromyographic (EMG) biofeedback, cognitive-behavioral therapy, or conservative medical treatment. At posttreatment, improvements were noted in all three treatment groups, with the biofeedback group displaying the most substantial change. At the 6- and 24-month follow-up, only the biofeedback group maintained significant reductions in pain severity, interference, affective distress, pain-related use of the health care system, stress-related reactivity of the affected muscles, and an increase in active coping self-statements. Treatment outcome was predicted by chronicity and treatment-specific variables. Analysis of attrition showed a significant effect for therapist and extent of somatic pathology. Results suggest that pain patients who suffer from musculoskeletal pain problems and display few physical disabilities may profit the most from short-term EMG biofeedback treatment.