Abnormal hemodynamic responses to exercise have been observed in diabetic subjects, but the pathogenesis and significance remain uncertain. We used maximal treadmill exercise to study 32 subjects with long-term insulin-dependent diabetes without clinical evidence of cardiac disease. Two of the 32 had occult ischemic heart disease revealed by stress electrocardiography and myocardial-perfusion scintigraphy and were excluded from subsequent analysis. In the remaining 30 subjects, we compared the responses to exercise of the 17 subjects with cardiac autonomic neuropathy diagnosed by noninvasive maneuvers (group 1) with the 13 without (group 2). At rest, the pressure-rate product (PRP) was higher in group 1 (114.0 +/- 5.7 vs. 95.9 +/- 5.3, P less than .05). With maximal exercise the increase in heart rate (44.6 +/- 4.8 vs. 79.0 +/- 5.4 beats/min, P less than .001), systolic blood pressure (36.8 +/- 5.9 vs. 55.0 +/- 5.8 mmHg, P = .02), and the PRP (102.0 +/- 7.3 vs. 182.0 +/- 8.2, P less than .001) were all lower in group 1 than in group 2, despite similar total treadmill times (631 +/- 47 vs. 587 +/- 40 s, P greater than .1). At each stage of exercise, the increase in heart rate and systolic blood pressure was lower in group 1 patients. The severity of cardiac autonomic neuropathy correlated inversely with the maximal increase in heart rate (r = -.68, P less than .001) and the PRP (r = -.58, P less than .005). Age, duration of diabetes, and the presence and severity of microvascular disease did not correlate with any of the hemodynamic parameters.(ABSTRACT TRUNCATED AT 250 WORDS)