In 1946 Berger and Bradley showed that mephenesin caused muscular relaxation and, in the same year, Stephen and Chandy (1947) reported that the intravenous administration of mephenesin to human subjects was usually accompanied by a feeling of euphoria and relaxation. The clinical trial of mephenesin in neurotic states associated with tension and anxiety was a natural corollary of these findings. Reports of its efficacy have been conflicting, but this is hardly surprising when a new drug is introduced for the treatment of a common disorder whose natural course may be either chronic or self-limiting, which is subject in a marked degree to apparently spontaneous fluctuations in intensity and on which the effects of suggestion are notorious. The assessment of therapy is made more difficult by the fact that there is no satisfactory objective method of measuring changes in the degree of anxiety; the investigator must either use tests the relevance of which many people will dispute or he must rely on clinical assessment, a subjective judgment very dependent on the skill and experience of the clinician. As Bradford Hill (1951) has stressed, clinical assessment can only be given full weight in a therapeutic trial when the possibility of enthusiasm or scepticism on the part of the clinician is eliminated. Such elimination of bias is best achieved by treating one group of patients with the drug and another group with a placebo, the clinician, nursing staff and patients not knowing to which group any particular patient has been assigned.