The question of when antidepressant drugs (AD) initiate significant clinical actions in depressed patients is still unsettled. Findings from early studies on whether there is a lag in the onset of therapeutic actions were in disagreement. More recent results with the selective serotonin reuptake inhibitors (SSRIs) and other new ADs indicate that clinical actions occur within the first 2 weeks. In this paper, evidence from efficacy studies with the ADs is reviewed and the methodologic and conceptual obstacles to achieving definitive results about the onset issue are analyzed. Depression, formerly viewed as a homogenous disorder, is now seen as heterogenous and multifaceted in structure. Such major structural components as anxiety and disturbed psychomotor functioning can be as significant to the core of the disorder as depressed mood itself. Further, the ADs have been shown to act initially on different facets of the clinical disorder which then result in multiple clinical actions, e.g., an initial reduction in anxiety followed by stimulation of motor activity. Data from the NIMH Collaborative Study of the Psychobiology of Depression are used to illustrate: (1) the componential structure of severe depressive disorder; (2) the sequence of change in the major behavioral components of the disorder associated with the tricyclic drugs; (3) the consequent "multiple" onsets of clinical actions; and (4) measurement of the clinical significance and visibility of the early behavioral changes. Recent results describing new behavioral and methodological approaches, the use of early clinical changes to predict outcome, and strategies for designing sound studies of onset are discussed.