Panic disorder is a chronic illness with only some degree of spontaneous recovery. It is not surprising therefore that discontinuation of an effective medical treatment may be followed by relapse. Therefore the timing and methodology of discontinuing that treatment are now recognized as essential facets of optimal clinical management. In addition to relapse, rebound and the withdrawal syndrome have been reported with many psychotropic agents, particularly with the benzodiazepines. This paper discusses data from three discontinuation studies with alprazolam i.e. the Phase I Cross-National Collaborative Panic discontinuation study after short-term treatment, the Phase I discontinuation study after long-term treatment, and data from the Montreal site of the Alprazolam SR discontinuation study. Phase I of the Cross-National Collaborative Study of Panic Disorder investigated the discontinuation of alprazolam in two populations. There was an intensive, placebo-controlled, time-limited study of discontinuation after short-term treatment (8 weeks) in the first population. For the second, there was a less rigorous open follow-up of patients who had been treated for 5-12 months with alprazolam. The dose-reduction regimen of alprazolam in both studies was approximately the same--a 1 mg decrease every 3-7 days. In the short-term treatment study, 109 patients were treated for 8 weeks, tapered for 4 weeks and observed for another 2 weeks post discontinuation. Significant relapse in the alprazolam-treated group occurred during discontinuation. Rebound of panic attacks occurred in 27% of patients given alprazolam, and distinct transient withdrawal syndrome occurred in 35%. Indicative of the withdrawal syndrome were confusion, clouded sensorium, heightened sensory perception, dysosmia, paresthesias, muscle cramps, muscle twitch, blurred vision, diarrhea, decreased appetite, and weight loss. The clinical course in the alprazolam-treated patients revealed a marked exacerbation of symptoms during the end of the tapering period and the first week without medication, which was followed by improvement during the second post-taper week. In the long-term treatment study, 142 patients were treated with alprazolam for periods ranging from 5 months to 1 year (mean, 27.5 weeks). In this naturalistic study, 76% of the patients reported improvement, 6.3% reported no change, and 10.6% reported that they were worse. During discontinuation, 12.8% of the 128 patients whose dosage was tapered reported some kind of nonspecified withdrawal symptoms. Of the 142 patients, 47.2% were able to taper their medication dosage and to discontinue treatment; 19.7% tapered but restarted alprazolam shortly after discontinuation; 33.1% were unable or unwilling to taper or discontinue alprazolam.(ABSTRACT TRUNCATED AT 400 WORDS)