Previous studies of the effect of angiotensin on myocardial contractility have yielded conflicting results. Possible reasons for the observed disparities include differences in techniques for measuring contractility, in species (dog, cat, and man), in myocardial state (normal or diseased), in preparation observed (heart-lung, isolated heart, papillary muscle, atrial myocardium, intact heart), and in dosage schedule. Moreover, there are no reported studies in the intact human heart, normal or diseased, in which contractility measurements are based on velocity-force relations. To resolve the conflict, left ventricular myocardial contractility was measured using the same expressions for the force-velocity relationship in all subjects. Studies were performed in five normal human subjects, six patients with cardiomyopathy, eight normal mongrel dogs, and six dogs with ischemic myocardial scarring, before and during angiotensin infusions in dosages producing 15--20-mm Hg increases of aortic diastolic pressure. Contractile element velocity at peak, dP/dt (Vce) and the Frank-Levinson contractility index (CyIx), which normalizes Vce for diastolic fiber length, decreased during angiotensin infusion in all groups. The mean decreases (11 to 19) per cent in Vce, 15 to 23 per cent in CyIx, SEM's 4-5 per cent) were significant (P values ranging from smaller than 0.05 to smaller 0.005) in the normal hearts of dogs and man and in the scarred canine hearts, in which preangiotensin Vce and CyIx were normal. In the cardiomyopathy group, in which contractility was depressed before angiotensin, the drug elicited a further decrease in Vce (mean fall 17 plus or minus 7 per cent, P smaller than 0.1) and CyIx (26 plus or minus 8 per cent, P smaller than 0.02). We conclude that, in the intact organism, with a normal myocardium or a diffuse or segmental myocardial disease, the administration of angiotensin results in a depression of contractility.