This study examined cardiovascular patterns following happiness, sadness, anger, fear, relaxation, and control imagery in 32 subjects while they were seated and while they exercised. Affective imagery was an effective strategy for inducing reliable patterns of systolic and diastolic blood pressure and heart rate associated with particular emotional states. Anger, rather than fear, produced the greatest overall increases in cardiovascular measures and was distinctly opposite from relaxation. Anger differed from fear and all other conditions in terms of greater increases in diastolic pressure following imagery and greater increases in heart rate and slower recovery of systolic pressure following exercise. Sadness was unique in that systolic pressure and heart rate were virtually as high when subjects were still as when they were actually moving. Furthermore, sadness was the one emotional state that seemed to interfere with the cardiovascular adjustments normally associated with exercise. Implications of these findings for current biobehavioral models of emotion, including the role that specific emotions may play in the pathogenesis and treatment of cardiovascular disease, are considered.