We monitored survival of 397 radiomarked juvenile American black ducks (Anas rubripes) distributed among Les Escoumins (n = 75) and Kamouraska, Quebec (n = 84), Amherst Point, Nova Scotia (n = 89), and a site on the Vermont-Quebec border (n = 149) during autumn 1990 and 1991. Eighty-six percent (215 of 250) of all confirmed mortalities during the study was from hunting; 72% of marked ducks were shot and retrieved and 14% were shot and unretrieved. We tested for differences in survival in relation to sex, body mass, year (1990-91, 1991-92), and among the 4 locations for each of 2 monitoring periods (early, EMP; late, LMP). With data from the EMP for Vermont-Quebec in 1990 and 1991, Les Escoumins in 1990, and Amherst Point in 1991, survival of hatching-year (HY) males and females did not differ (P = 0.357). For sexes combined for the EMP, survival of ducks was greater in 1991 than 1990 (P = 0.086), and differed among locations (P = 0.013). Survival (years combined) was greater at Amherst Point than at Kamouraska (P = 0.003) and Vermont-Quebec (P = 0.002) during the EMP. The highest survival rate at Amherst Point (0.545 ? 0.056 [SE]) was associated with the latest date (8 Oct) of season opening; the lowest survival rate (0.395 ? 0.043) was at the Vermont-Quebec border, where hunter numbers and activity were greatest. For the LMP, no interaction between years and locations was detected (P = 0.942), and no differences in survival existed between years (P = 0.102) and among locations (P = 0.349). No association was detected between body mass at capture and survival of combined males and females during the EMP (P = 0.572) or during the LMP (P = 0.965). When we censored hunting losses for combined years for each period, EMP or LMP, all survival estimates exceeded 0.800 (0.809-0.965). These data emphasize need for an improved harvest strategy for American black ducks in North America to allow for increases in breeding populations to achieve population goals.