Patients with social phobia who responded well to 6 months of open-label treatment with clonazepam were assigned to receive either continuation treatment (CT) with clonazepam for another 5 months, or to undergo discontinuation treatment (DT) using a clonazepam taper at the rate of 0.25 mg every 2 weeks, with double-blind placebo substitution. Clinical efficacy was compared between the CT and DT groups using three different social phobia scales. Benzodiazepine withdrawal symptoms were also measured. Relapse rates were 0 and 21.1% in the CT and DT groups, respectively. Subjects in the CT group generally showed a more favorable clinical response at midpoint and/or endpoint, although even in the DT group clinical response remained good. With respect to withdrawal symptoms, the rates were low in both groups (12.5% for CT and 27.7% for DT) with no real evidence suggesting significant withdrawal difficulties. At the end of 11 months of treatment with clonazepam, however, a more rapid withdrawal rate was associated with greater distress. This study offers preliminary evidence to suggest that continuation therapy with clonazepam in the treatment of social phobia is safe and effective, producing a somewhat greater clinical benefit than a slow-taper discontinuation regime. However, even in the DT group, withdrawal symptoms were not found to be a major problem. The study can be taken as supportive of benefit for longterm clonazepam treatment in social phobia, as well as being compatible with a reasonably good outcome after short-term treatment and slow taper.