In February 1983 a multifaceted study of the air‐ice‐ocean interaction was made in the southern Bering Sea. The study, MIZEX West, addressed a broad spectrum of physical problems related to the Bering Sea marginal ice zone (MIZ). As a part of that study an array of eight Argos‐tracked floes on a scale of 50 km provided information on the mesoscale behavior of the ice pack. Wind and current measuring platforms on two of the floes gave detailed information on the forces on individual floes which were compared with the larger‐scale motions. Under relatively steady northeast winds and with weak or negligible regional currents, the floes accelerated considerably as they crossed the MIZ. The array showed little distortion, even though it skirted around St. Matthew Island and changed trajectory direction by over 90° during the 12‐day study period. Similarly, individual floes remained within 20° of their original orientation, although their angular motions were erratic and often rapid. A baroclinicly forced acceleration of the wind over the MIZ contributed to floe acceleration. The relative currents at 2 and 6 m under the ice were highly variable and random in time, although well correlated in depth. One of the floes was much smoother than the other, and higher winds and currents were noted at the smoother floe. The motion of the floes reflect strong coupling to the currents at tidal and lower frequencies and low‐frequency (greater than 6 hours) response to the wind. Both floes drifted to the right of the mean wind by approximately 30° at about 4% of the wind speed at 3 m, and relative wind and current directions were within 20° of being colinear, traits that are consistent with free drift hypothesis. Using quadratic drag representations for wind and current stress, the near colinearity allowed calculation of the ratio of the air/ice to ice/water drag coefficients, Ca/Cw, relative to measurement heights of 3 m and −2 m. For the rough floe this ratio changed in mean from 0.06 to 0.2 in its movement to the MIZ. For the smooth floe the variation in the ratio was less, from 0.14 to 0.2. However, for the rough flow the angle between wind and current was generally less than 180°, and for the smooth flow it was greater. It is possible that another type of drag related to the bow effect of the floe pushing through the water can account for this observation. Finally, the motion of the ice floes and ice edge are compared by examination of satellite photos. Two regions of the ice pack—one thick with rafting and rubble, one thin and broken—showed a ratio of edge velocity to floe velocity of 0.64 and 0.43, respectively. Thinner ice overtakes the edge and melts more rapidly than the thicker ice. A simple thermodynamic model explains this observation.