The purpose of this study was to investigate the long-term maintenance of the efficacy of Lu AA21004 5 or 10 mg/day in the prevention of relapse in patients with generalized anxiety disorder (GAD). Patients (n = 687) with a primary diagnosis of GAD (DSM-IV criteria) and a baseline Hamilton Anxiety (HAM-A) total score of at least 20 underwent a 20-week, open-label Lu AA21004 treatment. In all, 459 patients responded and were randomized to 24-56 weeks of a double-blind treatment with Lu AA21004 (n = 229) or placebo (n = 230). The predefined primary efficacy endpoint was time to relapse (HAM-A total score ≥ 15) using a Cox model; the key secondary efficacy endpoint under multiplicity control was time to relapse for patients responding to treatment for at least 12 weeks. The primary analysis showed a statistically significant effect of Lu AA21004 relative to the placebo on the time to relapse of GAD, with a hazard ratio of 2.71 (P < 0.001). There was a statistically significant effect of Lu AA21004 in the stable responders (hazard ratio = 3.06, P < 0.001). Lu AA21004 was well tolerated, with withdrawal rates due to adverse events of 9% (open-label) and 3% (placebo) and 4% (Lu AA21004) in the double-blind period. In this study, Lu AA21004 5 or 10 mg/day was efficacious in preventing relapse and was well tolerated in the maintenance treatment of GAD.