Several psychological outcomes that accompany acute and chronic exercise have medical significance. Transient reductions in somatic tension and subjective anxiety appear most reliable. Symptom abatement in moderate depression can occur with chronic exercise in a manner comparable to psychotherapy and may offer a better prognosis in some instances. Other cognitive, behavioral, and perceptual events associated with exercise may assist in managing mental health, and exercise has been successfully used as a therapeutic adjunct in a variety of psychiatric disorders. Regular exercise may also complement treatments designed to manage aspects of coronary-prone behavior and psychoendocrine responsivity to mental stress. The lack of strict experimental control or effective placebo contrasts in most exercise studies precludes a convincing argument that exercise causes the psychological outcomes observed. Rather, expectancy of benefits, generalized treatment or attention effects, social reinforcement, and past history or selection bias represent likely alternatives. These competing explanations do not discount, however, that many individuals benefit in a clinically significant way. Exercise offers a low-cost alternative or adjunct with side effects that appear largely health-related. Although the effective psychological dosage or modality has not been quantified, current physiologic guidelines of the American College of Sports Medicine (large muscle rhythmic activity, for 20 to 60 minutes, 3 to 5 days per week at 60 to 80 per cent age-adjusted maximal heart rate), or a weekly caloric cost of 2000 kcal, should be effective with little medical risk. However, no evidence confirms that an increase in metabolic or psychoendocrine tolerance to exercise is necessary or sufficient for psychological outcomes to occur. Although biologic adaptations are known to follow exercise training and subside with diminished activity, there is currently no objective evidence that habitual exercise leads to dependence. If exercise has use in managing subjective or somatic symptoms, these may return during periods of exercise abstinence. Moreover, despite popular hypotheses concerning endorphins and biogenic amines, no direct relationships have yet been shown between exercise-induced mood swings and peripheral biochemical events. A proportion of habitual runners have reported acute episodes of euphoria-like states during or following exercise, but this remains a subjective and unpredictable event that may be related to psychophysiologic relaxation or acute changes in self-esteem.(ABSTRACT TRUNCATED AT 400 WORDS)