The MetaSwim (MS) metabolic cart can assess pulmonary gas exchange and ventilation in aquatic environments. The aims of this study were: 1) to determine the agreement between minute ventilation (V̇E), pulmonary oxygen uptake (V̇O2) and carbon dioxide output (V̇CO2) using the MS and Douglas Bag (DB) methods during flume swimming; 2) to assess the repeatability of these and other MS derived parameters. Sixteen trained swimmers completed a combined incremental and supramaximal verification cardiopulmonary swimming test to determine maximal V̇O2, two progressive intensity swimming tests during which MS and DB measurements were made (agreement protocol), and/or three-four constant velocity submaximal swimming tests during which only the MS was used (repeatability protocol). Agreement was determined using limits of agreement (LoA), bias, random error and 95% confidence intervals with systematic bias assessed using paired samples t-tests. Within-trial and between trial repeatability were determined using the coefficient of variation (CV) and the repeatability coefficient (CR). Where data were heteroscedastic, LoA and CR were log-transformed, anti-logged and displayed as ratios. MS underestimated peak V̇O2 and V̇CO2 (<0.39 L.min-1) and V̇E (9.08 L.min-1), while submaximal values varied between 2-5% for CV and ±1.09-1.22 for ratio CR. The test re-test CV during constant velocity swimming for V̇E, tidal volume, breathing frequency, V̇O2, V̇CO2, and end-tidal pressures of O2 and CO2 was < 9% (ratio CR of ±1.09-1.34). Thus, the MS and DB cannot be used interchangeably. Whether the MS is suitable for evaluating ventilatory and pulmonary responses in swimming will depend upon the size of effect required.