A group of patients suffering from anxiety, as assessed by general practitioners and psychologists using research criteria for generalised anxiety, were treated with either diazepam or placebo double blind for six weeks. This active treatment period was preceded by a one week single blind placebo "wash in" period and followed by a two week single blind placebo "wash out" period. The results suggest that diazepam can produce rebound anxiety and withdrawal symptoms when used in moderate doses and for what has previously been regarded as a safe length of time. If replicated these results have implications for the therapeutic use of benzodiazepines.