Removal of bird carcasses has been advocated for management of Clostridium botulinum outbreaks on lakes in North America because a reduction in density of toxin-laden maggots produced within bird carcasses is assumed to enhance survival of healthy birds. This inverse relationship between carcass density and survival has been reported in controlled studies with Mallards (Anas platyrhynchos) but has not been verified in wild ducks during naturally occurring botulism outbreaks. Therefore, we radio-marked 204 molting Mallards on seven lakes in western Canada during July–August 1999–2000, and monitored their survival daily for 30 days. Carcass searches were conducted simultaneously at 90 matched locations for freshly dead and randomly selected live radio-marked Mallards. Carcass density (carcasses/ha) averaged about two times greater at dead than at live duck locations (x̄=12.4, SE=1.2 vs. x̄=5.0, SE=0.7). Predicted risk of mortality increased rapidly with carcass density (case-control logistic regression: model-averaged β_density=0.167, unconditional SE=0.062). Mallards exposed to 5–11 and >11 carcasses/ha were 3.5 and 13 times more likely to die, respectively, than were Mallards inhabiting carcass-free areas. Mortality risk was more closely related to density of maggot-laden carcasses than to maggot-free carcass densities. Our results are consistent with the assumption that reducing carcass density could enhance survival. However, we caution that survival rates may remain low on lakes in which areas with high carcass densities persist due to incomplete carcass removal.