This multicentre study was conducted to evaluate the efficacy and consequences of progressive or abrupt withdrawal of clobazam in the treatment of Generalized Anxiety Disorder in a double blind study in comparison to lorazepam and buspirone. 128 outpatients suffering from Generalized Anxiety Disorder according to DMS III criteria were included in the study and treated for three weeks. They were randomly divided into 4 groups: group 1: 32 patients receiving clobazam, abruptly withdrawn and replaced by a placebo; group 2: 29 patients receiving clobazam with progressive withdrawal over 3 weeks, clobazam being replaced by a placebo; group 3: 33 patients receiving lorazepam with progressive withdrawal over 3 weeks, lorazepam being replaced by a placebo; group 4: 34 patients receiving buspirone, abruptly withdrawn and replaced by a placebo. The dosages were increased progressively during the first week of treatment. At the end of this time, the patients received either 30 mg clobazam or 30 mg buspirone or 3 mg lorazepam daily. After the first week, the Hamilton Anxiety Rating Scale (HARS) showed a significant improvement in clobazam and lorazepam groups but not in buspirone group. All the drugs were equally effective after three weeks of treatment. The anti-anxiety activity persisted after withdrawal of the studied drug in the 4 groups, without any signs of rebound anxiety or withdrawal syndrome. No clinically relevant differences were found between the 4 groups regarding safety. The side-effects reported were mainly drowsiness in clobazam and lorazepam groups, nausea and headache in buspirone group. In conclusion, clobazam like lorazepam improved anxiety more quickly than buspirone; after 3 weeks of therapy, efficacy was comparable with the 3 drugs and persisted after treatment discontinuation.