The validity of 2 electrothermal atomic absorption spectrometric methods for determination of selenium in foods and diets was tested. By using 0.5% Ni(II) as a matrix modifier to prevent selenium losses during the ashing step, it was shown that selenium can be determined in samples containing ≥1 μg Se/g dry wt without organic extraction. The mean recovery tested, using NBS Bovine Liver, was 98%; recovery of added inorganic selenium in Bovine Liver matrix was 100%. In addition, this method gave values closest to the median value of all participating laboratories using hydride generation AAS or the spectrof luorometric method in a collaborative study on high selenium wheat, flour, and toast samples. For samples with concentrations μg Se/g dry wt, separation of selenium from interfering Fe and P ions by organic extraction was necessary. Using inorganic ,5 Se in meat and human milk matrixes, an ammonium pyrrolidine dithiocarbamate-methyl isobutyl ketone- extraction system with added Cu(II) as a matrix modifier yielded the best extraction recoveries, 97 and %%, respectively. Accuracy and precision of the method were tested using several official and unofficial biological standard materials. The mean accuracy was within 4% of the certified or best values of the standard materials and the day-to-day variation was 9%. The Se/Fe or Se/P interference limits proved to be low enough not to affect selenium determinations in practically all foods or diets. The practical detection limit of the method was 3 ng Se/g dry wt for 1.0 g dry wt samples. Approximately 20 duplicate determinations can be performed per day. In terms of sensitivity, accuracy, precision, and sample throughput, the tested method compares favorably with the best methods available.