This study examined the accuracy and reliability of the MedGraphics VO2000 (VO2000) portable metabolic system and the ParvoMedics TrueOne 2400 (TrueOne 2400) metabolic cart against the criterion Douglas bag (DB) method. Ten healthy males (age 20 +/- 1.7 years) had their gas exchange variables measured at rest and during cycling at 50, 100, 150, 200, and 250 W. Each stage was 10-12 min. For half of the stage gas exchange was measured with the DB and TrueOne 2400 simultaneously and for the other half of the stage gas exchange was measured with the VO2000. The testing was performed on two separate days and the order in which the equipment was used in each stage was randomized. Reliability between days for V (E) (CV 7.3-8.8%) was similar among devices, however, for VO2, and VCO2 the VO2000 (CV 14.2-15.8%) was less reliable compared to the DB (CV 5.3-6.0%) and TrueOne 2400 (CV 4.7-5.7%). The TrueOne 2400 was not significantly different from the DB at rest or any work rate for V (E), VO2, or VCO2 (P > or = 0.05). The VO2000 was significantly different from the DB for V (E) at 50-100 W, VO2 at rest and 100-250 W, and VCO2 at rest and 200-250 W (all, P < 0.05). The TrueOne 2400 provides accurate and reliable results for the measurement of gas exchange variables. The VO2000 portable metabolic system was less reliable for measuring VO2 and VCO2 and generally overestimates VO2 at most cycling work rates. Further research is needed to confirm the results found with the VO2000.