In 10 controls and 43 patients with coronary artery disease (CAD) left ventricular (LV) diastolic pressure-volume (P-V) curves were obtained from biplane ventriculograms and simultaneous high fidelity pressure measurement at rest and during bicycle exercise. During exercise ventriculography 20 patients had angina pectoris, and 16 patients were asymptomatic. At rest there were no akinetic segments in 28 patients, and at least one akinetic segment was found in 15 patients. Shifts in the diastolic P-V relationship with exercise were quantitated from the constants a and b of the linear log P-V relationship. In the control group a and b did not change significantly, but in all CAD groups a significant decrease in a and a significant increase in b were observed during exercise. While no patient with angina had an unchanged diastolic P-V relationship, as many as 12 patients had significant P-V shifts in the absence of angina. A similar correlation was found for the diastolic P-V alterations and the exercise ECG. Fourteen patients without any ST-segment change during exercise showed significant P-V shifts, while no patient with signs of ischaemia in the ECG had an unchanged P-V curve. In another 20 patients with CAD the relative contribution of the Frank-Starling mechanism, diastolic compliance and the pericardium to the filling pressure rise during exercise was analyzed. Left ventricular and right atrial pressures--as an index of pericardial pressure--were measured simultaneously during rest and exercise ventriculogram. This was done when filling pressures exceeded 30 mmHg or when angina pectoris occurred.(ABSTRACT TRUNCATED AT 250 WORDS)