Abstract Seventeen patients suffering from exertional angina pectoris according to their records, and confirmed by exercise testing, have been subjected to a 40-week study (including a 13-week pretrial period) for assessing the effects of placebo, isosorbide dinitrate (ISD) 5 mg q.i.d. orally, and alprenolol (ALP) (an adrenergic β-blocker) 100 mg q.i.d. Variables measured were nitroglycerin consumption, scale rating of subjective symptoms, casual exercise testing, and step counter readings of daily activity. These variables were used to calculate a “severity index” and a “total response index”. ISD was found to have some, possibly a transient, beneficial effect, which was related to a decrease in systolic BP. The effect of ALP was significantly better in 14 of 17 patients responding to the drug, both with regard to nitroglycerin consumption, severity index, total response index and exercise testing. Placebo tablets reduced nitroglycerin consumption by about 25% over 10 weeks, but the severity of angina was not influenced. Side-effects were few on all drugs. Only headache was reported significantly more often during ISD therapy. The methodology of testing treatment efficacy in angina over prolonged periods is discussed, and the severity index is compared with conventional variables such as attack counting, subjective rating and casual exercise testing. The severity index correlates well with the total response index. It is concluded that the severity index, although theoretically the most suitable variable, is in practice difficult to determine, and that a total response index probably is as sensitive, and easier to obtain on the basis of conventional variables.