We evaluated a new exercise-testing system (Beckman Horizon MMC), incorporating a microprocessor that controls the acquisition of data, corrects for time delays, applies calibration factors, ensures quality control, and presents results in a variety of formats. Precision of measurements of ventilation (VE) and mixed expired gas concentrations was high. In steady-state exercise (n = 100) VO2 was measured with a precision (+/- SD) of 66 ml/min (4.3%), (r = 0.991); there was a small (4.62%) systematic underestimation of VCO2, but precision was comparable with VO2, with SD being 67 ml/min (4.55%) (r = 0.993). Good agreement was obtained between measurements made in progressive incremental exercise in healthy subjects with correlation coefficients of 0.997 for VE, 0.995 for VO2, and 0.994 for VCO2. Agreement in patients with cardiorespiratory disorders (n = 10) was similar, except in three patients in whom a variable pattern of breathing limited strict comparisons. Comparison with a breath-by-breath analysis system (n = 5) showed that rapid changes in VE, VCO2, and VO2 were followed accurately; the half time for a change in VO2 was not systematically different between the two systems (SD, 3.34 s, r = 0.951). The incorporation of microprocessor-controlled calibration procedures, which are simple to carry out frequently, was judged to be an important feature of this system.