The purpose of this study was to evaluate the reliability of the VmaxST portable metabolic measurement system. Forty-five healthy adults (age = 25.7 +/- 5.9 yr; height = 171.8 +/- 9.1 cm; weight = 69.6 +/- 12.8 kg; VO2peak) = 40.7 ml/kg/min; percent fat = 21.7 +/- 11.0) performed two separate and identical exercise routines on different days consisting of treadmill walking at 2.0 mph (53.6 m/min), 3.0 mph (80.5 m/min), and 4.0 mph (107.3 m/min) and running at 6.0 mph (160.9 m/min). VE and gas exchange were measured continuously breath-to-breath. A random effects model on log-transformed data yielded coefficients of variation (CV) and intraclass correlation coefficients (ICC) for VO2 and VE of 5.2 - 7.6 %, and 0.77 - 0.92, respectively, for all walking and running trials. For VCO2, CVs were higher (10 - 12 %) and ICCs lower (0.70 - 0.81). Ordinary least squares regression between the individual difference scores and the individual mean scores for VE, VO2 and VCO2, respectively, indicated no systematic bias (all p > 0.05). Bland-Altman analysis also illustrated no systematic bias between repeated measurements. The VmaxST provides reliable measurements of VO2 and VE during walking and running eliciting VE and VO2 at least up to approximately 56 and 2.2 l/min, respectively. The system appears to be less reliable for measuring VCO2.