The purpose of this study was to assess the validity and reliability of a Cosmed K4b2 portable telemetric gas analysis system. Twelve physically fit males performed a treadmill running session consisting of an easy 10 min run, a hard 3 min run and a 1 min sprint (with rest periods of 10 min separating each run), on four separate occasions. Sessions were identical with the exception of the apparatus used to measure O2. During two (test-retest) sessions a Cosmed K4b2 portable gas analysis system was used; in another, a laboratory metabolic cart, and in one session, both systems were used to measure O2 simultaneously. Comparison of Cosmed K4b2 and metabolic cart measurements in isolation revealed significantly (p<0.05) increased values of O2, CO2, FE CO2 (except FE CO2 at 10min) and lower values of FE O2 for each run duration by the Cosmed system. Linear regression equations to predict metabolic cart results from Cosmed values were, respectively; cart O2 = 0.926 ' Cosmed O2 ' 0.227 (r2 = 0.84) and cart CO2 = 1.057 ' Cosmed CO2 ' 0.606 (r2 = 0.92). Bland 'Altman plots and comparison of the test-retest Cosmed measurements revealed that the K4b2 system showed good repeatability of measurement for measures of E, O2 and CO2, particularly for 10 and 3 min tests (ICC = 0.7 ' 0.9, p<0.05). In conclusion, the Cosmed K4b2 portable gas analysis system recorded consistently higher O2 and CO2 measurements in comparison to a metabolic cart, however, satisfactory test-retest reliability of the system was demonstrated.