Although North American wood ducks (Aix sponsa) are well-studied throughout their range, researchers know little about demographic and environmental factors influencing survival of ducklings and broods, which is necessary information for population management. We studied radiomarked female and duckling wood ducks that used nest boxes and palustrine wetlands at Noxubee National Wildlife Refuge (NNWR) in Mississippi, USA, in 1996-1999, and riverine wetlands of the Tennessee-Tombigbee Rivers and Waterway (TTRW) system in Alabama in 1998-1999. We estimated survival of ducklings and broods and evaluated potentially important predictors of duckling survival, including age and body mass of brood-rearing females, hatch date of ducklings, duckling mass, brood size at nest departure, inter-day travel distance by ducklings, site and habitat use, and daily minimum air temperature and precipitation. At NNWR, survival of 300 radiomarked ducklings ranged from 0.15 (95% CI = 0.04-0.27) to 0.24 (95% CI = 0.13-0.38) and was 0.21 (95% CI = 0.15-0.28) for 1996-1999. Our overall estimate of brood survival was 0.64 (n = 91; 95% CI = 0.54-0.73). At TTRW, survival of 129 radiomarked ducklings was 0.29 in 1998 (95% CI = 0.20-0.41) and 1999 (95% CI = 0.13-0.45) and was 0.29 (95% CI = 0.20-0.40) for 1998-1999. Our overall estimate of brood survival was 0.71 (n = 38; 95% CI = 0.56-0.85). At NNWR, models that included all predictor variables best explained variation in duckling survival. Akaike weight (wi) for the best model was 0.81, suggesting it was superior to other models (<0.01 ??? wi ???0.18). We detected 4 competing models for duckling survival at TTRW. Inter-day distance traveled by ducklings was important as this variable appeared in all 4 models; duckling survival was positively related to this variable. Patterns of habitat-related survival were similar at both study areas. Ducklings in broods that used scrub-shrub habitats disjunct from wetlands containing aggregations of nest boxes had greater survival probabilities than birds remaining in wetlands with such nest structures. Managers may increase local wood duck recruitment by promoting availability of suitable brood habitats (e.g., scrub-shrub wetlands) without aggregations of nest boxes that may attract predators and by dispersing nest boxes amid or adjacent to these habitats. We did not determine an optimal density of nest boxes relative to local or regional population goals, which remains important research and conservation needs.