Four distinct sub-ice mesoscale features were observed during the 1983 Marginal Ice Zone Experiment in Fram Strait. The first one was an anticyclonic eddy located over the Yermak Plateau and strongly correlated with bottom topography, however, it appears not to have been topographically generated. The remaining features were observed along the ice edge of the East Greenland Current. Two were closely related in that one was a filament of Atlantic Water that was apparently entrained by a warm core cyclonic eddy located northwest of the Molloy Deep. The last one could not be shown conclusively as a westward extension of a branch of Atlantic Water or an isolated cyclonic eddy. Simplified estimates show that, on average, each of the three features had enough available heat energy residing within the first 30 m of their cores to melt over 600 km2 of 2-m-thick ice. Other estimates of potential melting and retreat rates of the marginal ice zone are calculated on the basis of various utilization depths from which available heat energy can be drawn. Further consideration is also given to the possibility that these types of features may also act as ''destructive preconditioners'' of sea ice since it is unlikely that complete melting would occur because of their relative movements. Eddies and intrusions significantly increase the amount of available temperature-salinity (T-S) space within which mixing can be accomplished. These admixtures produced by direct interaction of Atlantic Water or Atlantic Intermediate Water and the Polar Waters of the East Greenland Current further provide a very effective mechanism which transports salt to deeper levels through salt fingering and double diffusion. Calculations of heat, mass, and salt fluxes for the observed features show that the heat transported by the ice edge eddies was significant in comparison with a depth comparable flux of the East Greenland Current, while mass and salt fluxes were roughly 7 times less. ¿American Geophysical Union 1987