The reproducibility of physical capacity expressed as cumulative work (CW) and work pulse (WP) as well as haemodynamic response to maximal exercise was evaluated by two different principles for increasing work load during a maximal bicycle ergometer exercise test: (i) stepwise increment (SI) of 50 Watts (W) every fourth minute; (ii) continuous increment (CI) of 10 W at the end of every minute, both with a starting load of 50 W. Forty apparently healthy men aged 29-69 (mean 47) years performed two tests with a mean interval of 17 (3-43) days. Group A: 10 men who did SI twice; group B: 10 who repeated CI; group C: 10 men who performed SI and CI alternatively; group D: 10 men who did SI in the morning vs afternoon. Error % (percentage of the absolute difference of the mean of two values) was 3.9 for CW in group A vs 7.8 in group B (p less than 0.05). The error % of heart rates at submaximal work levels were lower with SI than CI. Mean CW in group C was 182.3 +/- 65.1 kJ for CI and 154.8 +/- 48.8 kJ for SI (p less than 0.01), whereas the mean exercise time was 1206 +/- 297 s and 1089 +/- 197 s respectively (p less than 0.01). Group D subjects achieved 142.7 +/- 49.4 kJ in the morning and 134.4 +/- 48.8 kJ in the afternoon (p less than 0.05). No significant differences were noted for WP in groups A-D. A better reproducibility for CW was found for SI, whereas both tests had similar reproducibility for heart rate and blood pressure response during maximal exercise.