The effect of diazepam and barbital in the treatment of delirium tremens and other acute conditions related to alcohol abuse was evaluated in a double-blind trial. 91 patients participated in the study, 44 in the diazepam group, 47 in the barbital group. The choice of diazepam rather than chlordiazepoxide was motivated by its major anticonvulsive properties. Barbital was given by the oral route, diazepam as intramuscular injections. Different ways of drug administration to patients with delirium tremens are discussed. It is concluded that the two different ways used in the study probably did not have a noteworthy influence on the results. All patients were excluded who had taken psychoactive drugs before admission. Nevertheless a considerable part of the patients had diazepam, but not barbital, in the blood before treatment was initiated. This may give support to the use of barbital as a "special purpose drug" in the treatment of these conditions. The patients were divided into three diagnostic categories, according to the severity of the clinical condition. No difference between the two drugs tested was found in the milder conditions, but barbital was found superior to diazepam in the treatment of fully developed delirium tremens.