Objective: To determine which variables predicted long-term outcome after cognitive behavioral therapy (CBT) for chronic fatigue syndrome (CFS). Methods: A cohort of 511 CFS patients from four different CBT for CFS studies, i.e. two cohort studies and two RCT's. Before treatment, all patients fulfilled the 2003 US CDC criteria for CFS and treated with CBT, were assessed at long-term follow-up, up to 10 years after end of treatment. We tried to predict fatigue severity and physical functioning at follow-up with demographics, cognitive-behavioral perpetuating factors, and CFS characteristics as predictors in linear regression analyses. Logistic regression analysis was used to explore significant predictors of fatigue scores within normal limits at long-term follow-up. Results: Lower fatigue severity at long-term follow-up was predicted by a shorter duration of CFS symptoms and lower fatigue levels at baseline, and lower frustration in response to fatigue and lower fatigue levels directly post-treatment. Fatigue scores within normal limits at follow-up was predicted by lower fatigue severity and lower levels of frustration in response to fatigue, both assessed directly post-treatment. Better physical functioning at follow-up was predicted by higher sense of control over fatigue, better physical functioning at post-treatment, and being younger at baseline. In some of the additional analysis pain at baseline also predicted physical functioning at follow-up. Conclusion: The finding that lower fatigue severity and higher physical functioning at long-term follow-up were positively associated with its outcomes at post-treatment underline the importance of fully maximizing the positive effects of CBT for the sustainment of outcomes. Furthermore, augmenting sense of control and starting treatment sooner after diagnosing CFS could positively influence long-term outcome. Interventions aimed at pain management deserve more attention in research.