ABSTRACT Ice accumulation measurements made on Middleton Island in the Gulf of Alaska at a 10 meter elevation during the winter of 1975-1976 and on St. Paul Island is the Bering Sea during the fall of 1976 and the winter of 1976-1977 at 10, 20, and 30 meter elevations are reported along with meteorological conditions prevalent before, during and after icing events. The data obtained is used in an attempt to establish preliminary methods for design recommendations for offshore atmospheric superstructure icing on derricks, flare booms, antennas, quarters and superstructures on an offshore platform located in sub arctic or Arctic regions. CLIMATOLOGICAL CONSIDERATIONS Konishi and Saito noted that a two year cycle seemed evident in the wind, ice and temperature patterns in the Bering Sea for the period 1960 to 1971. The range of temperature variation (based on May values) was of the order of 3.6 °F centered around 35.6 °F, with odd years being warmer and even years being cooler. Exceptions were found in 1965 and 1971. During these periods the northerly winds also prevailed as they did in the "low" temperature even years. Evidently, what is reflected are differences in the position of the Aleutian low and the North Pacific high. Generally storms enter the Bering Sea from the southwest or south, sometimes from Kamchatka on the west or from the Gulf of Alaska on the southeast. Many die as they encounter ice fields in winter, others recur across the Alaska peninsula into the Gulf of Alaska and some move north-eastward up the Kuskokwin and Yukon River (Figure 1). Studies of ship icing show that maximum icing generally occurs in the rear of low pressure areas, during N, NW and Winds; however, a sub-maximum of icing events may also occur in the forward part of a low with NE or E winds. According to Borisenkov and Pchelko (1972X 57% of the 442 reported cases of ship icing in the Bering Sea occurred in the rear of a low pressure area, while 32% of the events were reported in the foreword part of the low; all other cases accounted for 11%. The icing period defined by these reports is from December through March, with corresponding frequency of ship icing of 20%. This is not so striking a low pressure system dependence as is found elsewhere. In the sea of Japan for example, 93% of the icing events occur in the rear of a low pressure system. It is even more interesting to note that according to the locations of these icing events reported by Borisenkov and Pchelko (1972), almost all were clustered about the Pribilofs with the balance occurring to the SE, as far as western Bristol Bay. Only about 11 events occurred outside this area. Thus, these figures are truly representative of the study area, the SE Bering Sea. Certainly these data are biased towards the location of known fishing grounds, since the vessels reporting tend to concentrate in these areas.