Experiments are reviewed in conscious dogs instrumented with ultrasonic crystals for measuring systolic wall thickening and myocardial blood flow (microsphere technique), and studied under a variety of conditions during acute or chronic circumflex coronary stenosis. With acute progressive stenosis, a nearly linear relation was found between normalized subendocardial flow and function. There were no consistent ST segment changes in the body surface electrocardiogram if wall thickening was less than 25%, although changes occurred in the subendocardial electrogram. The regional flow-function relation was then examined during exercise with various degrees of coronary stenosis; the mean flow-function relation was shifted to the right of the resting relation, but when subendocardial flow was expressed per beat and function normalized as a fraction of that in normal wall the relationship was superimposable upon the resting relation, suggesting that ischemia is absolute (not relative) in the subendocardium during steady state conditions. When the normal increase of wall thickening during exercise was prevented by mild coronary stenosis ischemia could not be clearly detected, but when function during exercise averaged 20% below the resting value, subendocardial blood flow changes and other evidence of ischemia were readily apparent, indicating the sensitivity of wall motion for detecting subcritical coronary stenosis. The regional flow-function relation was also studied in chronic single vessel coronary artery stenosis (ameroid constrictor), in which function was normal at rest (collateral development) but exercise produced large decreases in both regional flow and function. At matched levels of treadmill exercise, a calcium channel blocker (diltiazem) together with beta blockade (atenolol) produced an additive effect that was greater than with either drug alone, with substantial increases in both subendocardial flow and regional function. Finally, studies showing reversible post-reperfusion dysfunction after 15 minutes or two hours of coronary occlusion are reviewed. Partial ischemia for 5-hours, followed by reperfusion also produced regional dysfunction which persisted for at least 3 days but reverted to normal by one week, with little or no histologic damage of the free wall. These studies on regional flow and function during ischemia may carry implications for a number of important clinical phenomena.