In order to investigate the reproducibility of maximum progressive exercise we studied 6 healthy volunteers twice a day for 3 nonconsecutive days. The average within-subject coefficients of variation for the maximal ventilatory and heart rate (HR) responses ranged from 3.8% for HR to 12% for minute ventilation (Ve) and anaerobic threshold. Three-way analysis of variance revealed no significant variations in exercise parameters for the group as a whole except for the carbon dioxide output (VCO2) and respiratory exchange ratio (R) which was significantly greater in the afternoon (p less than 0.05). This was attributed to dietary effects of carbohydrate loading upon VCO2 and was associated in 1 subject with a significant increase in Ve (p less than 0.01). In another subject, morning values of Ve were consistently and significantly (p less than 0.05) greater than the afternoon values which could only be attributed to increased effort as indicated by an increased oxygen uptake. Tests of resting pulmonary mechanics (FEV1, FVC, FRC, RAW) measured before each exercise procedure showed no significant diurnal or day-to-day variations. Results indicate that while the maximal responses to progressive exercise are generally reproducible and the first exercise procedure can usually be considered representative, diurnal variations in R, VCO2 and Ve may occur which can be best avoided on repeated testing by exercising subjects at a standardized time of day.