To evaluate and identify baseline characteristics of the adolescent patients included in two previous randomized controlled trials (RCTs) that may predict a clinically significant outcome after treatment of temporomandibular disorders (TMD) with an occlusal appliance (OA) or relaxation training (RT) in a clinical sample of adolescents. This study combined two patient samples from the earlier RCTs for a total of 167 adolescents with frequent TMD pain (once a week or more often), diagnosed according to the Research Diagnostic Criteria for TMD. They were treated with OA, RT, or received information only (control). Outcome (response to treatment vs nonresponse) was assessed using four measures: the Patient Global Impression of Change (PGIC), pain intensity rated on a numeric rating scale (NRS), pain frequency levels, and pain severity levels prospectively recorded in a pain diary. Predictors of outcome were evaluated posttreatment for the whole sample and at 6 months follow-up for participants from the first trial. Associations and differences between groups obtained in the bivariate analyses were further examined in subsequent multivariate logistic regression analyses. At posttreatment, treatment condition (OA being more effective than RT/control), gender (boys being more responsive than girls), arthralgia (predicting lower response), lower levels of somatic complaints (predicting better response), and shorter TMD pain history (predicting better response) emerged as significant predictors of a clinical response. At 6-month follow-up, lower consumption of analgesics and shorter TMD pain history emerged as significant predictors of treatment outcome, while treatment condition approached significance after multivariate analysis. This study revealed that treatment condition and gender were the most consistent predictors of a clinically significant outcome across outcome measures in a clinical sample of adolescents with TMD. Treatment with OA reduced TMD pain in the adolescents.