The problems discussed include the difficulty of defining appropriate criteria of clinical benefit", the difficulty inherent in the satisfactory conduct of prolonged large scale multi-centre trials; the problem of changes in concurrent therapy, particularly in the present context of the introduction of coronary care units; the problem of the extent to which laboratory control should be mandatory in trial design and the difficulties involved in extrapolating from a result obtained in a meticulously designed controlled trial, executed by dedicated investigators in college hospitals to routine use in general public hospitals.