The reproducibility and accuracy of measurements by the Beckman Metabolic Measurment Cart and by an established manual method of the ventilatory responses and aerobic requirements at rest and during symptom-limited exercise were evaluated in eight normal subjects. Each person performed four exercise tests. Both methods showed excellent reproducibility of minute ventilation, oxygen uptake (VO2), carbon dioxide production (VCO2), and respiratory exchange ratio (RER). The respective correlations for the metabolic measurement cart were r = .99, .99, .99, .93 and for the manual method r = .98, .99, .99, .93. Reproducibility of heart rate and blood pressure responses during exercise was good, with correlation coefficients of .95 and .91 for values obtained during tests with the metabolic measurement cart and .94 and .89 with the manual method. Each variable at maximal exercise showed good reproducibility with no significant difference for either method. Comparison between the pooled metabolic cart measurement and the pooled manual method measurements of variables during the third minute of each stage of exercise showed excellent agreement (r = .97, .98, .99, .93 for minute ventilation, VO2, VCO2, and RER, respectively). The means of differences observed between the two methods were small (.69, .04, .03 L/min and .00 for each variable, respectively), although VO2and VCO2showed significant differences (P < .05). Comparison of variables between the pooled results for the two methods at maximal exercise showed high agreement. We conclude that the metabolic measurement cart provides reproducible and accurate measurements of respiratory variables and on-line results that are an advantage over the manual method. It also saves technician time.