Eighty-two patients with social phobia were randomly assigned to cognitive therapy or double-blind drug treatment with moclobemide (300-600 mg per day) or placebo. After 15 weeks of active treatment, a 2-month treatment-free follow-up as well as a 15-month naturalistic follow-up were scheduled. At post-test, cognitive therapy was significantly superior to moclobemide, but not to placebo, on a composite social phobia measure. At 2-month follow-up, cognitive therapy was superior to both moclobemide and placebo. Treatment gains were maintained in cognitive therapy during 2-month and 15-month follow-up, however, most patients (66%) who completed the study needed additional cognitive and/or behavioural treatment. Moclobemide proved not superior to placebo at post-test as well as at 2-month follow-up. As most patients from the medication conditions were treated with cognitive and/or behavioural treatment during the naturalistic follow-up period, at the 15-month assessment no between-group differential effects were found to remain. These results indicate that cognitive therapy is an effective treatment for social phobia in both the short and long term. As the results from previous studies on the efficacy of moclobemide in social phobia were inconclusive, our data tip the scales in the direction that there is no place for moclobemide in the treatment of social phobia.