Waterfowl biologists need reliable predictors of brood and duckling survival to accurately estimate recruitment rates. We examined 30-day survival rates of gadwall (Anas strepera) broods (1992-1994) and ducklings (1990-1994) in eastern North Dakota, USA, during years when water conditions ranged from extremely dry to extremely wet. We evaluated effects of several variables on brood survival: (1) percent of seasonal wetland basins containing water, (2) occurrence of rain on the current or 2 previous exposure days, (3) minimum ambient temperature averaged over the current and 2 previous exposure days, (4) hatch date, (5) duckling age, and (6) brood size. Only 9 of 58 radiomarked females lost their entire broods; Kaplan-Meier 30-day survival rate for broods was 0.84. Brood size (adjusted for hatch date) was a better brood-survival predictor than were any of the environmental variables. Risk of total brood loss decreased by 24% for each additional duckling in the brood. We monitored fates of 212 radiomarked ducklings from 94 broods. Daily risk of death for ducklings was twice as high when seasonal ponds were scarce as when ponds were abundant. Duckling survival rate was lower during the first 7 days of life whether it rained or not; survival rate was greater for 8- to 30-day-old ducklings, but was reduced by rain. Thirty-day duckling survival was greatest when minimum daily temperatures exceeded 10°C and no rain occurred. We attributed 86% of 87 deaths of radiomarked ducklings to predation; American mink (Mustela vison) accounted for ≥68% of the 40 deaths for which predator type could be ascertained. Despite apparent resilience of gadwall populations during drought, our study documented a positive effect of seasonal wetland availability on gadwall duckling survival. Management efforts to improve recruitment will be more effective in years when most seasonal basins contain water.