The relation between end-tidal carbon dioxide tension (PETCO2) measured by infrared analysis and arterial carbon dioxide tension (PaCO2) during exercise was systematically examined in five healthy adults at two power outputs (25 and 50% VO2max) and at three frequencies of breathing (15, 30, and 45 breaths/min). PETCO2-PaCO2 varied between -2.5 and +9.1 Torr, was inversely related to the frequency of breathing (r = 0.475), and directly related to tidal volume (VT; r = 0.791) and CO2 output (r = 0.627). An equation was obtained by multiple regression analysis, to predict PaCO2 from PETCO2: PaCO2 = 5.5 +0.90 PETCO2 -0.0021 VT (r = 0.915). The equation was applied to measurements of PETCO2 obtained in two previous studies in 10 subjects in which PaCO2 had been measured, and was found to predict PaCO2 to within 1.04 Torr (+/- SD) for PaCO2 between 25 and 58 Torr (n = 56; r = 0.962). The effect of the response characteristics of the CO2 analyzer on the measurement of PETCO2 was also systematically examined by comparison with a fast-responding respiratory mass spectrometer.