Information concerning the validity and reliability of commercial on-line gas analysis systems is limited. The aim of this study was to provide a comparison of the validity and reliability of three on-line systems (Oxycon Alpha, Oxycon Pro and Pulmolab EX670) with that of Douglas bags. Two separate studies were conducted. In study 1, the three gas analysis systems were compared with Douglas bags using a metabolic simulator over four increases in ventilation. In study 2, ten subjects were split into equal groups exercising at 100 W or 150 W for 85 min on three separate occasions. Each system was used twice per visit. Study 1 demonstrated that the Oxycon Alpha and Douglas bags produced similar respiratory values over all levels of ventilation. The Oxycon Pro tended to slightly overestimate mean expiratory flow (V(E) ), oxygen uptake ( VO2), carbon dioxide production (VCO2) and respiratory exchange ratio (RER) at the higher ventilations. The Pulmolab produced large overestimations at all ventilations for and RER (up to 26.3% away from expectations), whilst values for, and were slightly underestimated at higher ventilations (up to 7.5% from expectations). The results of study 2 support the findings of study 1, with the Oxycon Pro and Oxycon Alpha producing similar results compared to Douglas bags for, and RER. The coefficients of variation for and measured using Douglas bags, Oxycon Pro and Oxycon Alpha were 3.3-5.1%, 4.7-7.0% and 4.5-6.3%, respectively, whilst that for the Pulmolab was highly variable (26.8-45.8%). The exercise study showed the Oxycon Pro and Oxycon Alpha to be both valid and reliable on-line systems for the measurement of parameters of respiration, at least at workloads up to 150 W.