Population levels of emperor geese (Chen canagica) in Alaska in 1993 were about half that estimated in the 1960s. Survival information is necessary for managers to decide how to best enhance recovery of this species to former levels. We calculated seasonal and annual estimates of emperor goose survival from resightings of neck-collared birds. Geese were neck collared in 1988-90 on their breeding grounds in the Yukon-Kuskokwim Delta, Alaska, and resighted each spring and fall, 1988-92, at staging areas on the Alaska Peninsula. Adult monthly survival rates during overwinter periods (1 Oct-30 Apr) were not different (P = 0.281) among years (Ŝ'= 0.940, SE = 0.009), whereas monthly rates of oversummer (1 May-30 Sep) survival showed annual variation (P = 0.048). However, we constrained oversummer survival to a single estimate of 0.980 (0.010). Monthly survival estimates for juveniles during their first overwinter period did not vary among years (P = 0.999) and was 0.710 (0.018). Subsequent monthly survival for juveniles was 0.943 (0.010), similar to that for adults. We developed an adjustment procedure to account for philopatric behavior of geese and this enabled us to use data for postbanding (1 Aug-30 Sep) periods. Survival estimates were low compared with those for other goose species, particularly for juveniles. We addressed collar loss and heterogeneity in resighting probabilities and felt their contribution to potential model bias was insignificant. Annual survival among adults (Ŝ' = 0.631, SE = 0.023) was not different (P = 0.709) from that observed during 1982-85 (Petersen 1992). The similarity in survival rates in these studies suggests that harvest regimes did not differ between the 2 periods. This suggests that continued subsistence harvest has contributed to persistent low population levels in emperor geese.