a ecological planning and toxicology, inc., 5010 SW Hout St., Corvallis, OR 97333-9540 ABSTRACT Aquatic birds are exposed to selenium through their diet by ingesting aquatic invertebrates that have accumulated selenium from water and the food chain. However, dietary composition is highly variable among species, over time, and across sites, making it difficult to provide accurate estimations of dietary exposure for particular species at specific locations. Selenium accumulates in the egg, resulting in embryo malformation, embryonic death, and decreased survival of juveniles. If the relationship between egg concentration and these reproductive parameters can be defined with sufficient certainty, then risk assessments can be performed through analysis of egg selenium concentrations. Other researchers have proposed egg toxicity thresholds that lead to conclusions of widespread selenium toxicoses in waterbirds. However, we believe these values are overly conservative and that it is unlikely that selenium is posing a significant risk to wild birds in areas where the current water quality criterion is being met. Through the use of simple statistical models (logit, probit, and Weibull functions) we are able to express mortality and teratogenicity relationships for mallards (Anas platyrhynchos) in such a manner that the risk manager can be presented with information about the probability of reduced duckling survival if mean egg selenium (MES) concentrations are known. Data analysis indicates that the two endpoints (mortality * Corresponding author: Dr. Anne Fairbrother, ecological planning and toxicology, inc., 5010 SW Hout St., Corvallis, OR 97333-9540;Fairbrother, et al.and teratogenesis) cannot be distinguished statistically. Commonly used effects thresholds (the EC 10 and EC 20 ) correspond to 16 and 21 mg/kg dw MES using the most sensitive endpoint, chick mortality. Both of these values are higher than the 6 mg/kg dw level proposed by Skorupa (1998) who based his estimate on field-observational data potentially confounded by other environmental stress factors. The mortality and teratogenicity endpoints presented here relate selenium exposure to risk to individuals within a population and do not provide information about the probability of selenium causing changes in population growth rates, either at the local or regional levels.