In recent years, several new pharmaceutical agents have been developed for the treatment of affective disorders. The `second generation' antidepressants are said to offer significant advantages over the prototypical tricyclic agents. In the first part of this article the requirements to be come up to by clinical trials, which are intended to adequately characterize the therapeutic usefulness, are briefly reviewed. It is concluded, that methodological problems in clinical studied and inadequate reporting make it often difficult (or impossible) to judge their efficacy. This is especially true for newer antidepressant drugs, which have been studies less frequently. In the second part the author discusses the importance of those characteristics of cyclic antidepressants and the validity of those statements, which form the basis of the usually applied criteria of choice. It is concluded, that selectivity in catecholamine/indolamine reuptake inhibition does not offer any clinical advantages. The popular view of the specificity of some antidepressants for retardation and others for agitation is not supported by the available evidence. Criteria of choice may result from the anticholinergic effects and the potency to induce postural hypotension. In addition, the pharmacokinetic characteristics of cyclic antidepressants are dealt with. Although classical antidepressants are far from ideal in this respect, it is concluded that no relevant criteria of choice can be formulated. Second generation antidepressants do not show clearly superior pharmacokinetics, and plasma level monitoring offers the opportunity to correct for variation in the case of classical antidepressants.