We contrasted body condition, and age-specific reproduction and mortality between a growing population of sea otters (Enhydralutris) at Kodiak Island and a high-density near-equilibrium population at Amchitka Island, Alaska. We obtained data from marked individuals, population surveys, and collections of beach-cast carcasses. Mass:length ratios indicated that females (but not males) captured in 1992 at Amchitka were in poorer condition than those captured at Kodiak in 1986–1987. In 1993, the condition of females at Amchitka improved in apparent response to two factors: (1) an episodic influx of Pacific smooth lumpsuckers, Aptocyclus ventricocus, from the epi-pelagic zone, which otters consumed; and (2) an increase in the otters’ benthic invertebrate prey resulting from declining otter numbers. Reproductive rates varied with age (0.37 [CI=0.21 to 0.53] births female⁻¹ yr⁻¹ for 2–3-yr-olds, and 0.83 [CI=0.69 to 0.90] for females ≥4 yr old), and were similar at both areas. Weaning success (pups surviving to ≥120 d), in contrast, was almost 50% lower at Amchitka than at Kodiak and for females ≥4 yr of age was 0.52 (CI=0.38 to 0.66) vs 0.94 (CI=0.75 to 0.99), respectively. Sixty-two percent of the preweaning pup losses at Amchitka occurred within a month of parturition and 79% within two months. Postweaning survival was also low at Amchitka as only 18% of instrumented pups were known to be alive one year after mother-pup separation. Adult survival rates appeared similar at Amchitka and Kodiak. Factors affecting survival early in life thus are a primary demographic mechanism of population regulation in sea otters. By maintaining uniformly high reproductive rates over time and limiting investment in any particular reproductive event, sea otters can take advantage of unpredictable environmental changes favorable to pup survival. This strategy is consistent with predictions of “bet-hedging” life history models.