The accuracy of many microprocessor controlled exercise testing systems has not been established, particularly for testing children. A metabolic cart (Sensormedics S2900Z) was validated against a reference system (Douglas bag) in children. The systems were tested using 10 active, healthy children (age 11.6 ± 2.3 yrs) during physiological steady state at two exercise intensities, submaximal walking and running. Eight children performed a high ventilatory flow (HVF) protocol, running for 11 min. All 10 performed a low ventilatory flow (LVF) protocol, walking for 11 min. In both, gas samples were measured after 6 to 7 and 10 to 11 min into the Douglas bag, and after 8 to 9 min by the test system. There was minimal evidence of bias forat either ventilation rate. Greater variability was noted for FeCO2and. In conclusion, the Sensormedics S2900Z appears to be adequate for testing children.Key words: microprocessor, bias, ventilatory flows