Estimates of the DMS sea-air transfer velocity (kDMS) derived from direct flux measurements are poorly modeled by parameterizations based solely on wind speed and Schmidt number. DMS and CO2 flux measurements show kCO2 to be a stronger function of wind speed than kDMS. The NOAA/COARE gas flux parameterization, incorporating the bubble-mediated gas transfer theory of Woolf (1997), appears to do a better job reproducing the observations for both gases, illustrating the importance of trace gas solubility in sea-air exchange. The development of gas transfer parameterizations based on physical principles is still in its infancy, but recent advances in direct flux measurement methods provide an opportunity to evaluate the success of various modeling approaches for this critical geophysical process.