To evaluate the effects of different methods of detection, exercise modes, protocols, and reviewers on oxygen uptake (VO2) at the ventilatory threshold (ATge), 17 men with heart disease (mean age 59 +/- 6 years) and six healthy men (mean age 60 +/- 11 years) underwent six exercise tests on different days. Each subject performed three treadmill tests (Bruce, Balke, and ramp) and three bicycle ergometer tests (50 W/stage, 25 W/stage, and ramp) in random order. The ventilatory threshold was determined for each of the six exercise tests by three independent, blinded reviewers by means of graphic plots of three commonly used methods of determination: (1) changes in the ventilatory equivalents for VO2 and VCO2, (2) changes in end-tidal oxygen and carbon dioxide pressures, and (3) the intersection of the slope of VCO2 and VO2 (V slope). The largest variability in the ATge was observed with changes in the exercise protocol. The greatest absolute (ml/min) and percentage differences in oxygen uptake at the ATge as a result of changes in protocol, method of determination, and observers were 336 (36%), 125 (12%), and 70 (7%), respectively. The overall intraclass correlation coefficient for VO2 at the ATge among the three reviewers was 0.60 and among the three protocols was 0.85 (p less than 0.01). The V slope method of detection had consistently good agreement among reviewers and was least affected by the protocol. The variance in the ATge (excluding intersubject and error variance) accounted for by differences in protocol, method, and reviewer was 82%, 14%, and 4%, respectively.(ABSTRACT TRUNCATED AT 250 WORDS)