Twenty outpatients suffering from chronic anxiety states completed a trial of anxiolytic drug therapy. They each received five treatments: amylobarbitone sodium, chlordiazepoxide, diazepam, medazepam, and a placebo in flexible dosage in a fully balanced design, using doubleblind procedures. They were assessed on the Hamilton rating scale for anxiety after a period of two to four weeks on each of the treatments and they completed self-ratings every three days. The selfratings were composed of at least three prime symptoms chosen by the patient himself and phrased in his own words as well as ‘loss of appetite’ and ‘insomnia’. Neither placebo nor amylobarbitone produced any substantial improvement from pre-drug levels of anxiety but all three benzodiazepines produced significant and very similar decreases in both Hamilton ratings and self-ratings. These two measures also correlated highly. The majority of the patients continued with medazepam at the end of the formal trial.