We fed mallards (Anas platyrhynchos) diets supplemented with 0-, 1-, 2-, 4-, 8-, or 16-ppm selenium in the form of selenomethionine. We fed another group of mallards a diet containing 16-ppm selenium as selenocystine. Females fed the control diet produced a mean of 8.1 ducklings that survived to 6 days of age, which was significantly greater than the 4.6 young produced by females fed 8-ppm selenium as selenomethionine and the zero surviving young of females fed 16-ppm selenium as selenomethionine. Selenocystine did not impair reproduction. Diets containing 8- and 16-ppm selenium as selenomethionine caused malformations in 6.8 and 67.9%, respectively, of unhatched eggs compared with 0.6% for controls. The most common malformations were of eyes, bill, legs, and feet. Selenium did not affect the onset or frequency of egg laying, egg size, shell thickness, fertility of eggs, or sex ratio of ducklings. Reduced survival and growth occurred in ducklings hatched from groups whose parents had received 8- or 16-ppm selenium as selenomethionine, even though all ducklings were fed a control diet. Concentrations of selenium in eggs and liver of adults could be predicted from dietary concentrations. We conclude that the dietary threshold of selenium as selenomethionine necessary to impair reproduction is between 4 and 8 ppm. It is difficult to identify 1 level of selenium in eggs that will be diagnostic of reproductive impairment in the field because different chemical forms of selenium appear to have different toxicities in eggs. However, when eggs from a wild population contain .gtoreq. 1-ppm selenium on a wet-weight basis, reproductive impairment may be possible and should be evaluated in that population. At 5-ppm selenium in eggs, reproductive impairment is much more likely to occur.