In this study, follow-up results of cognitive-behaviour therapy and of a combination of cognitive-behaviour therapy with a serotonergic antidepressant were determined. The study also examined factors that can predict this treatment effect, both in the long term and in the short term. In addition, it investigated whether differential prediction is possible for cognitive-behaviour therapy vs. a combination of cognitive-behaviour therapy with a serotonergic antidepressant. A total of 99 patients were included in the study. Treatment lasted 16 weeks, and a naturalistic follow-up measurement was made 6 months later. Of the 70 patients who completed the treatment, follow-up information was available for 61 subjects. Significant time effects were found on all outcome measures at both post-treatment measurement and follow-up. No differences in efficacy were found between the treatment conditions. Effectiveness at post-treatment measurement appears to predict success at follow-up. However, 17 of the 45 non-responders at the post-treatment measurement had become responders by the follow-up. The severity of symptoms, motivation for treatment and the dimensional score on the PDQ-R for cluster A personality disorder appear to predict treatment outcome. No predictors were found that related specifically to cognitive-behaviour therapy or combined treatment. These results indicate that the effectiveness of cognitive-behaviour therapy or a combination of cognitive-behaviour therapy and fluvoxamine at the post-treatment measurement is maintained at follow-up. However, non-response at post-treatment does not always imply non-response at follow-up. Patients with more severe symptoms need a longer period of therapy to become responders. Although predictors for treatment success were found, no evidence was found to determine the choice of one of the treatment modalities.