Many exercise protocols are in use in clinical cardiology, but no single test is applicable to the wide range of patients' exercise capacity. A new protocol was devised that starts at a low workload and increases by 15% of the previous workload every minute. This is the first protocol to be based on exponential rather than linear increments in workload. The new protocol (standardised exponential exercise protocol, STEEP) is suitable for use on either a treadmill or a bicycle ergometer. This protocol was compared with standard protocols in 30 healthy male volunteers, each of whom performed four exercise tests: the STEEP treadmill and bicycle protocols, a modified Bruce treadmill protocol, and a 20 W/min bicycle protocol. During the two STEEP tests the subjects' oxygen consumption rose gradually and exponentially and there was close agreement between the bicycle and the treadmill protocols. A higher proportion of subjects completed the treadmill than the bicycle protocol. Submaximal heart rates were slightly higher during the bicycle test. The STEEP protocol took less time than the modified Bruce treadmill protocol, which tended to produce plateaux in oxygen consumption during the early stages. The 20 W/min bicycle protocol does not take account of subjects' body weight and consequently produced large intersubject variability in oxygen consumption. The STEEP protocol can be used on either a treadmill or a bicycle ergometer and it should be suitable for a wide range of patients.