The flux of CO2 between the atmosphere and the ocean is often estimated as the air-sea gas concentration difference multiplied by the gas transfer velocity (K660). The first order driver for K660 over the ocean is wind through its influence on near surface hydrodynamics. However, field observations have shown substantial variability in the wind speed dependencies of K660. In this study we measured K660 with the eddy covariance technique during a ~ 11,000 km long Southern Ocean transect. In parallel, we made a novel measurement of the gas transfer efficiency (GTE) based on partial equilibration of CO2 using a Segmented Flow Coil Equilibrator system. GTE varied by 20% during the transect, was distinct in different water masses, and related to K660. At a moderate wind speed of 7 m s-1, K660 associated with high GTE exceeded K660 with low GTE by 30% in the mean. The sensitivity of K660 towards GTE was stronger at lower wind speeds and weaker at higher wind speeds. Naturally-occurring organics in seawater, some of which are surface active, may be the cause of the variability in GTE and in K660. Neglecting these variations could result in biases in the computed air-sea CO2 fluxes.