Wave slopes, from a pitch and roll buoy, and wind velocity components, from a platform-mounted anemometer, was sampled simultaneously during a 1984 study on the Grand Banks of Newfoundland. The principal direction of wave slope was highly correlated with the 10-m wind direction. Residual errors in direction were large only for wind speeds less than 5.0 m/s. When these estimates were removed from the record, there was an 8.0¿ rms error in the estimated wind direction. The upwind slope variance correlated at 0.90 with the 10-m wind speed. A quadratic regression of wind speed to upwind slope variance resulted in an rms error of 1.9 m/s. A rotary spectral analysis showed an extremely high coherence between the slope estimated wind components and the directly measured surface wind components for periods greater than 18.0 hours. It is suggested that the technique could be applied to surface following pitch and roll buoys to estimate surface wind speeds. ¿ American Geophysical Union 1989