A new method was used to study the effect of a single dose of propranolol on the QT intervals during exercise in 11 normal volunteers. They exercised maximally on a bicycle ergometer and repeated the test after taking propranolol (40 mg) by mouth two hours before. Electrocardiograms were continuously recorded on magnetic tape and the cardiac cycle length (RR interval) and the QT interval were measured every five seconds by a computer aided method. The RR-QT data from each test during the exercise phase were analysed by an exponential formula, QT = A - B x exp (-k x RR) and by Bazett's formula, QT = K x square root of (RR). Three reference QT intervals, QTc1, QTc2, and QTc3, estimated at RR = 400, 700, and 1000 ms respectively from the regression curves of both formulas were compared. The exponential formula, which consistently gave a better fit with the data, showed that propranolol had a biphasic action on the QT intervals during exercise. It significantly prolonged the mean (SD) interval at longer cycle lengths (from 287 (27) to 305 (18) ms at RR = 1000 ms and shortened it at shorter cycle lengths (from 198 (14) to 179 (16) ms at RR = 400 ms). In contrast, Bazett's formula did not show any significant effect when the same raw data were used. The exponential formula can be adapted to study other interventions or conditions that affect QT intervals.