A feeding study with mallard ducks (Anas platyrhynchos) was conducted during March to July 1988 in Laurel, Maryland (USA), to identify diagnostic criteria for selenium toxicosis in birds. One-year-old male mallards in groups of 21 were fed diets containing 0, 10, 20, 40, or 80 parts per million (ppm) selenium, as seleno-DL-methionine, for 16 weeks. All ducks receiving 80 ppm died. Ducks receiving 40 or 80 ppm selenium consumed less feed than ducks in the other treatment groups. Body weights of ducks receiving 40 or 80 ppm selenium declined during the study. The post-breeding molt was delayed in ducks receiving 40 ppm; most ducks receiving 80 ppm selenium died prior to the onset of molt. At necropsy, numerous abnormalities were observed in ducks that died but only a small number of abnormalities were observed in ducks surviving to the end of the study in the 40 ppm group. Weights of the heart, spleen, and pancreas were mostly lower and weights of the kidney were higher for ducks dying during the study than for euthanized ducks. Liver weights were unaffected. Selenium accumulated in soft tissues approximately in proportion to dietary concentrations. Selenium concentrations in tissues of all ducks that died were different from those of surviving ducks in the 0, 10, and 20 ppm groups, but were not different from those of surviving ducks in the 40 ppm group. Proposed diagnostic criteria for fatal chronic selenosis were derived from body weight, macroscopic abnormalities, organ weights, and concentrations of selenium in the liver. Proposed diagnostic criteria for non-fatal chronic selenosis were derived from body weight, plumage condition, macroscopic abnormalities, concentrations of selenium in the liver, reproductive failure, and alterations of blood and tissue chemistries. Lead or dioxin poisoning have diagnostic criteria most similar to selenium toxicosis.