The results from three separate treatment studies on adolescents (N = 108) aged 16-18 years with predominantly recurrent tension (n = 85), and migraine (n = 23) headaches are presented along with information on predictors of short-term outcome. A standardized therapist-or home-based relaxation training program was compared to attention-control approaches in which information or problem-discussion procedures were provided for the students. These treatment conditions were both contrasted with self-monitoring procedures. All students were assessed and treated in a school setting and may therefore, be regarded as representing a nonclinical replication series. The outcome results indicated that the majority (53%) of students, in particular those with tension headache, had improved after both forms of relaxation training, which were clearly superior to attention-placebo control and self-monitoring contrasts. Furthermore, it was found that treatment condition, baseline headache severity and students' satisfaction with their home life were the strongest predictors of overall outcome, accounting for 32% of the total variance. For students provided with relaxation training, a set of four parameters, again including baseline headache severity, in addition to students' level of somatic symptoms, school satisfaction and disease within their family, emerged as strong predictors of outcome, explaining 44% of the outcome variance. The results are discussed in view of similar outcome and prediction studies in the field of pediatric and adult recurrent headache.