Isotopic evaporation models, such as the Craig-Gordon model, rely on the description of non-equilibrium fractionation factors that are, in general, poorly constrained. To date, only a few gradient-diffusion type measurements have been performed in ocean settings to test the validity of the commonly used non-equilibrium fractionation factor parametrizations for ocean evaporation. In this work we present six months of water vapor isotopic observations collected from a meteorological tower located in the northwest Atlantic Ocean (Bermuda) with the objective of estimating the best non-equilibrium fractionation factors (k, ‰) for ocean evaporation and their dependency on wind speed. Gradient-diffusion measurements are sensitive enough to resolve non-equilibrium fractionation factors during evaporation and provide mean values of k18= 5.2±0.6 ‰ and k2= 4.3±3.4 ‰. In this study, we furthermore evaluate the relationship between k and 10-m wind speed over the ocean. Such relationship is expected from current evaporation theory and from laboratory experiments made in the 1970s, but observational evidence is lacking. We show that (i) sensitivity of k to wind speed is small, in the order of -0.16 to 0.20 ‰ s/m for k18, and (ii) there is no empirical evidence for the presence of a discontinuity between smooth and rough wind speed regime during isotopic fractionation, as proposed in earlier studies. Instead, k18 monotonically decreases within our observed wind speed range [0 – 10 m/s]. Implications for using such k values in modelling ocean vapor d-excess are briefly discussed.