A study was conducted to assess the feasibility of substituting diazepam for pentobarbital and phenobarbital in sedative-hypnotic tolerance testing and withdrawal of tolerant patients. Nineteen patients potentially dependent on sedative-hypnotics were tolerance tested with either diazepam or pentobarbital and, if tolerant, were withdrawn with diazepam or phenobarbital, respectively. Seventy-eight subjects previously described in the literature with drug dependencies similar to the diazepam and barbiturate groups provided a basis for comparing the diazepam group to untreated patients to determine whether diazepam was able to prevent abstinence symptoms. Withdrawal was successfully completed in all patients. The expected and observed incidence of seizures, delirium, and minor withdrawal for the barbiturate group and seizures and delirium for the diazepam group did not differ significantly, although a large type II error compromised these conclusions. A significant difference was seen between the observed and expected incidence of minor withdrawal symptoms in the diazepam-withdrawn patients.