The therapeutic effectiveness of sulpiride on various types of neurosis was compared with that of chlordiazepoxide on a double-blind basis. Global improvement, effectiveness on various types of neurosis, and side effects were studied for a period of two weeks. The subjects consisted of 41 males and 32 females. Neurosis including borderline case and vegetative dystonia was divided into eight different subtypes comprising borderline, neurasthenic state, hypochondria, obsessive neurosis or phobia, depressive neurosis, anxiety neurosis, vegetative dystonia, and others. A newly devised matched-pair method of comparison was employed to achieve even distribution of the eight subtypes of neurosis between the two drug groups. As a result 40 patients fell in the sulpiride group and 33 patients in the chlordiazepoxide group. The backgrounds of the paired patients matched closely. The daily dose was uniformly 150 mg for sulpiride and 30 mg for chlordiazepoxide in the first week but was raised (or lowered in some cases) to 225 mg and 45 mg, respectively, in the second week according to severity. The rate of global improvement was 79% for the sulpiride group and 90% for chlordiazepoxide group, and the difference did not reach statistical significance. Improvement by manifestation (13 symptom items) and type of neurosis also matched. Side effects occurred at a rate of 28% (sulpiride group) and 30% (chlordiazepoxide group), and also matched closely in incidence and variety. The authors concluded that sulpiride in appropriate doses is useful in the treatment of neurosis without causing extrapyramidal side effects.