Four exercise test protocols (Bruce, Balke, Ellestad and Steep) were compared in 16 patients with proven coronary artery disease in demonstrating the anti-anginal effects of sublingual glyceryl trinitrate in a randomization double-blind trial. Glyceryl trinitrate significantly improved the time, heart rate and rate pressure product to peak exercise, onset of angina and 1 mm ST segment depression in all four protocols (P < 0.05) (except rate pressure product to angina in the Balke protocol). The increase in exercise time was greatest for the Balke protocol at peak exercise (188.1 +/- 187.1) (mean +/- SD in s), at onset of angina (251.9 +/- 247.1) and at 1 mm ST depression (233.6 +/- 243.8), followed by the Steep and Bruce protocols, and was lowest for the Ellestad protocol 41.9 +/- 42.4, 96.5 +/- 65.8, 82.6 +/- 74.0, respectively. Increase in time to peak exercise with glyceryl trinitrate was significantly greater for the Balke protocol in comparison with the other three protocols and for the Bruce and Steep protocols when compared to the Ellestad protocol. Time to 1 mm ST depression with treatment was significantly greater on the Balke and Bruce protocols than the Ellestad protocol, and to onset of angina for the Balke compared to other three protocols. There were no significant differences between the Bruce and Steep protocols for any of the endpoints. The magnitude of treatment effect in the different protocols was accompanied by correspondingly greater inter-patient variability such that no protocol was more, or less, sensitive than another in detecting treatment effect. Changes in heart rate and rate pressure product with treatment were generally similar between the different protocols. In conclusion, a protocol with small and frequent increments (Balke), although able to show greater increase in exercise duration with glyceryl trinitrate than more aggressive protocols, is no more sensitive at detecting treatment effect.