With the increased use of oxygen consumption measurements in clinical and sporting studies, measurement variability has become more important to both the paediatric clinician and the sports scientist. In this study we assessed the reproducibility of cardiorespiratory measurements during submaximal and maximal running in children. Ten healthy, physically active boys (mean(s.d.) age 10.7(0.71) years) performed two submaximal and two maximal running tests within a 4 week period. The submaximal protocol consisted of three 6 minute runs at 7.2, 8.0 and 8.8 km/h. Every attempt was made to minimize the sources of non-biological variability at each testing session. During submaximal exercise, oxygen consumption (VO2), heart rate (HR) and fractional utilization appeared to be the most reliable measures accounting for over two-thirds of the total variation (coefficients of reliability (CR) of 68%, 94% and 82% respectively). Ventilation (Ve) and respiratory exchange ratio (RER) proved to be only moderately reliable accounting for less than half of the total variation (CR 50% and 45% respectively). At maximal exercise, VO2, Ve and time to exhaustion were most reliable, accounting for approximately two-thirds of the total variation (CR 65%, 63% and 63% respectively). Within this test environment, a two visit submaximal assessment was capable of estimating VO2 with a standard error of +/- 1.25 ml/kg/min. Similarly, for maximal testing a two visit assessment estimated peak VO2 with a standard error of +/- 2.28 ml/kg/min. On the evidence of these results a two visit assessment for submaximal and maximal exercise testing seems adequate to estimate the stability of submaximal cardiorespiratory measures and peak VO2 in healthy, normally active boys.