The U.S. Marine Mammal Protection Act (MMPA) and International Agreement on the Conservation of Polar Bears mandate that boundaries and sizes of polar bear (Ursus maritimus) populations be known so they can be managed at optimum sustainable levels. However, data to estimate polar bear numbers for the Chukchi/Bering Sea and Beaufort Sea populations in Alaska are limited. We evaluated aerial line transect methodology for assessing the size of these Alaskan polar bear populations during pilot studies in spring 1987 and summer 1994. In April and May 1987 we flew 12.239 km of transect lines in the northern Bering, Chukchi, and western Beaufort seas. In June 1994 we flew 6.244 km of transect lines in a primary survey unit using a helicopter, and 5,701 km of transect lines in a secondary survey unit using a fixed-wing aircraft in the Beaufort Sea. We examined visibility bias in aerial transect surveys, double counts by independent observers, single-season mark-resight methods, the suitability of using polar bear sign to stratify the study area, and adaptive sampling methods. Fifteen polar bear groups were observed during the 1987 study. Probability of detecting bears decreased with increasing perpendicular distance from the transect line, and probability of detecting polar bear groups likely increased with increasing group size. We estimated population density in high density areas to be 446 km²/bear. In 1994, 15 polar bear groups were observed by independent front and rear seat observers on transect lines in the primary survey unit. Density estimates ranged from 284 km²/bear to 197 km²/bear depending on the model selected. Low polar bear numbers scattered over large areas of polar ice in 1987 indicated that spring is a poor time to conduct aerial surveys. Based on the 1994 survey we determined that ship-based helicopter or land-based fixed-wing aerial surveys conducted at the ice-edge in late summer-early fall may produce robust density estimates for polar bear populations in the Chukchi/Bering and Beaufort seas.