Anemometer‐measured winds for the period 5–13 March 1994 were used to study the coherence of observed and forecast coastal winds along the mid‐Labrador shelf. The reliability of these variables in predicting the response of the ocean and ice to wind forcing is an important issue for ice forecasting in this area. Two anemometer‐equipped 2‐m ice beacons were deployed on pack ice north of Wolf Island and a third beacon was deployed on Grady Island. The results indicate that due to the influence of local topography, 10‐m winds observed at the meteorological station in Cartwright, Labrador provide a poor estimate (r2 = 0.2) of wind conditions over the offshore sea‐ice. In contrast, the σ = 1 level (∼10 m) winds from the Canadian Meteorological Centre's Regional Finite Element (RFE) model provided a better correlation with anemometer beacon winds (0.90 for the 6‐hour forecast down to 0.45 at 36 hours). However, the RFE model overestimates the magnitude of the winds by 10–40%. The response of the ocean and ice cover to wind forcing was measured by an ocean bottom‐mounted acoustic Doppler current proþler (ADCP). Relative to the 2‐m beacon winds, the ice moved at 2.5% of wind magnitude and turned 0.6° to the left of the wind. The ocean response decreased with depth until it reached a constant value of 0.9% of the wind speed. The turning angle increased from 0.3° to the right of the wind at 3.5 m to 50° at the lowest level measured by the ADCP (73 m depth). Approximately 57% of the variance in the ocean currents at 3 m below the surface can be attributed to the 2‐m winds; at 73 m the explained variance decreases to 27%.