Meta-analysis revealed that in studies evaluating behavioral treatments for tension headaches, the treatment outcome has varied with the client samples (e.g., age, gender, referral source) that have been used but not with the treatment procedures (e.g., type of behavioral intervention, length of treatment, whether or not efforts were made to facilitate transfer of training) or the research designs (e.g., internal validity, explicitness of diagnostic criteria) that have been used. Mean client age proved the best predictor of treatment outcome, accounting for 30% of the outcome variance following behavior therapy. Significantly poorer outcomes have also been reported in recent studies than were reported in early studies. These findings suggest that outcomes obtained with behavioral interventions have been less dependent upon the treatment variables that have been the primary focus of research attention than upon characteristics of client samples and behavioral interventions may be less effective in reducing headache activity than has previously been assumed.