To define the clinical and angiographic features of the syndrome of spasm of angiographically normal coronary arteries, 77 patients with spasm and fixed angiographically normal coronary arteries, 77 patients with spasm and fixed coronary stenosis equal to or greater than 50% (group A) were compared with 35 patients with spasm and normal or minimally diseased coronary arteries (group B). Statistically significant differences between groups A and B were as follows: the incidence of rest angina (50.6% vs 85.7%; p less than 0.01) and mixed angina (32.5% vs 5.7%; p less than 0.01); the appearance of ST segment depression (53.3% vs 16%; p less than 0.01) and no electrocardiographic changes during stress test (35.6% vs 76%; p less than 0.01); and the tendency for arteriographically documented spasm to be focal (87.5% vs 71.4%; p less than 0.05) and to affect only one vessel (76.6% vs 57.1%; p less than 0.05). No differences were found between groups A and B in major coronary risk factors, history of previous myocardial infarction, electrocardiographic abnormalities at rest or during pain episodes, or arteries affected by spasm. Thus, angina appearing exclusively at rest is the main clinical feature of spasm of normal coronary arteries. The electrocardiogram, whether at rest or during pain episodes, has no value for predicting the existence of underlying coronary lesions, whereas stress testing does. Spasm of normal arteries tends to be more diffuse than that superimposed on organic lesions and to affect more than one artery, suggesting different mechanisms in the genesis of both types of spasm.