Dimensions of health beliefs (perceived risk of behavior and benefit of behavior change), social support (family and others' support for change), and self-efficacy (magnitude and strength) were examined in 215 patients undergoing a prospective trial of health promotion in a primary care medical practice. Discriminant analyses were performed to evaluate how well these dimensions predicted motivation for change and lifestyle behavior change. These relationships were examined for six lifestyle areas: cigarette smoking, dealing with stress, amount and type of food eaten, use of seat belts, and exercise habits. The analyses demonstrated a statistically and clinically significant prediction of motivation by one or more health belief and self-efficacy dimensions for most lifestyle areas. The strongest single predictors were perceived benefits and self-efficacy strength, which were each significant predictors of motivation in four lifestyle areas (P < 0.05). Support dimensions, as measured, were not shown to have predictive value in most areas. Behavior change was poorly predicted by beliefs, support, and self-efficacy for most lifestyle areas. However, adding motivation to the discriminant function equation resulted in significant predictions in all six lifestyle areas (P < 0.05), with an average correct classification rate of 71%. This finding strongly suggests that motivation is a very important intervening variable when evaluating health promotion and resulting behavior change. © 1991.