Meltwater input from sea ice forms a buoyancy source for the upper ocean which creates a strong density gradient in both horizontal and vertical directions. If, in particular, the ocean density front is formed in the Bering sea during winter, the frontal dynamics are influenced by local shelf/slope processes. further, ice is advected across the front by the wind, thereby altering the heat flux to the ice and leading to an increase in the freshwater buoyancy flux to the ocean. Hence, th~ surface manifestation of the front is governed by ice position. In this thesis a detailed study of ice and ocean parameters in such a system is presented using data from the 1982-3 winter Season in the Bering sea. Particular attention is given to the results from MIZEX-West (1983), an intensive mid-winter study. Modelling of the physical processes involved in the development of the meltwater front follows two directions; firstly the buoyancy input to small scale fronts formed in the summer marginal ice zone is considered and secondly, an hierarchy of numerical models of ice and ocean dynamics are employed. Results are also reported of fieldwork carried out in the East Greenland current (MIZEX-84, LANCE cruise) using a novel, medium resolution, portable CTD system to measure upper ocean density gradients from floe edges and small boats. However, during these experiments conditions were not ideal for the development of meltwater fronts analogous to those found in the Bering Sea. Additionally, one-dimensional and two-layer quasi-steady ocean models coupled to an ice cover are discussed. These models proved useful tools in our understanding of the air-ice-ocean exchanges and frontal adjustment processes within the more complex system. More detailed modelling studies were undertaken using a two-dimensional, coupled ice-ocean model focusing on the interactive thermodynamic forcing during ice ablation and ice accretion. The thermal and salinity fluxes in the coupling were related to the ice growth calculated by a thermodynamic ice model similar to those is employed by larger scale climate studies. Hence, over short time scales the ice growth in leads and open water that may occur after a change in external forcing conditions is not well represented. The functional form of the internal stress in the ice momentum equation is investigated with the model. When the amplitude of the internal stress decreases by several orders of magnitude in a few grid points the model was unable to sustain such a gradient and was liable to generate numerical instabilities. If however, the near discontinuity in the internal stress at the ice edge is treated like a moving shock wave (after Roed and O'Brien, 1983) then the ice compactness maintains a coherent ice edge under conditions simulating the passage of a storm.