Nifedipine is an effective antianginal agent, but its efficacy in patients with angina refractory to maximally tolerated conventional therapy has not been well studied. We reviewed the experience using nifedipine in an unblinded manner in 716 patients with refractory angina, all of whom underwent cardiac catheterization. Patients were treated with nifedipine when maximally tolerated conventional therapy was inadequate to control angina. Patients were divided into three mutually exclusive clinical groups based on the presumed pathophysiologic mechanism responsible for angina. Group I consisted of 389 patients with Prinzmetal's angina and coronary vasospasm documented by the observation of spontaneous angina with ST segment elevation and/or vasospasm observed during coronary angiography. Group II was composed of 292 patients with "mixed angina," defined as those patients who exhibited evidence of both classic exertional angina as well as possible superimposed coronary vasospasm. None of these patients had documented coronary vasospasm or ST segment elevation with angina. Group III included 35 patients with classic stable exertional angina, without rest pain or ST segment elevation associated with episodes of ischemia. Angina frequency and nitroglycerin use were compared on conventional therapy before and after the addition of nifedipine. Mean duration of nifedipine therapy was 6.5 months. The addition of nifedipine (median dose 60 mg/day, range 10 to 200 mg) significantly decreased the mean frequency of angina attacks/week in group I from 14.4 to 3.0 (p < 0.0001), in group II from 19.9 to 5.9 (p < 0.0001), and in group III from 11.3 to 7.1 (p < 0.03). Complete prevention of angina was most frequent in patients with documented vasospasm (42% of group I patients), intermediate in those clinically suspected of but not proven to have vasospasm (20% of group II patients), and least frequent in patients with classic exertional angina alone (3% of group III patients) (p < 0.001). In 78% of the 716 patients the weekly angina rate decreased by ≥50% of baseline values obtained during maximally tolerated conventional therapy, but the degree of improvement was significantly better in patients with either suspected or documented vasospasm. Treatment with nifedipine was associated with an increase of angina frequency in 13% to 29% of the 716 patients; this increase was most frequently observed in those with no evidence of vasospasm (group III). Nifedipine efficacy did not vary on the basis of the presence or absence of fixed obstructive coronary disease. These results suggest that nifedipine is efficacious for patients with angina refractory to maximally tolerated conventional therapy, and that efficacy may be greatest when coronary vasospasm is also present.