A stratified random sampling procedure involving a helicopter census of square-mile (2.6 km2) quadrats in a 648 mi2 (1,678 km2) pilot area in the AOSERP study area was continued in February 1977. The census produced a weighted mean estimate of 0.49 moose/mi2 (0.19/km2), or a study area estimate of 320 moose ± 29 percent. This was not significantly different from the population of 363± 30 percent estimated in 1976. The 1977 moose population of 320 moose was contained in an estimated 196 groups distributed on 23 percent of the square-mile study area quadrats at the time of census. Analysis of front and back observer data indicated a visibility bias in the 1977 census of approximately 50 percent. Adequate visibility bias models are required to provide biologically meaningful and statistically reliable estimates on any future moose censuses of the area.