OBJECTIVE To describe the routine management of patients with chronic stable angina by GPs in Northern Ireland and the factors which they perceived affected the success of medical therapy. METHODS A questionnaire survey of all general practitioners in Northern Ireland (n = 962). METHODS A survey conducted collaboratively by the Departments of Public Health Medicine in each of the four Health Boards in the province. Total population served, 1.5 million. METHODS The relationship between the perceived reasons for medical treatment failure and the stated referral and prescribing practice of the GPs. RESULTS A total of 541 GPs replied; the response rate was 56%. The two most important reasons given for the perceived failure of medical therapy were (i) underlying disease progression and (ii) an adverse patient lifestyle such as smoking or obesity (cited as of primary importance by (i) 264 and (ii) 225 doctors respectively). The ranking differed significantly according to the doctor's propensity to prescribe triple therapy, with those doctors in the highest tertile of this distribution being less likely to cite the patient's lifestyle as a primary reason for treatment failure (chi-squared = 6.7, d.f. = 2, P = 0.035) and more likely to cite underlying disease progression as a primary reason (chi-square = 7.0, d.f. = 2, p = 0.031). The overall ranking of the primary reasons for referral differed significantly according to the proportion of patients given a trial of triple therapy and to the doctor's propensity to refer. Doctors who had given a greater proportion of their patients at least a trial of triple therapy (in the highest tertile of the distribution) were more likely to cite the need for revascularisation assessment as the primary reason (chi-square = 12.5, d.f. = 2, P = 0.0019). On the other hand, the need for further advice on medical therapy was generally ranked higher by those doctors who had given fewer of their patients at least a trial of triple therapy (chi-square = 7.3, d.f. = 2, P = 0.027). GPs who had referred fewer of their new patients to hospital were more likely to be those doctors with fewer patients given at least a trial of triple therapy. Doctors with a greater percentage of their patients managed primarily by a hospital specialist tended to have more who had had a trial of triple therapy for their symptoms. CONCLUSIONS The results suggest the need for clearer definition for GPs of the place of revascularisation and of medical therapy for patients with stable angina.