The sensitivity of solid-phase radioimmunoassay for the measurement of staphylococcal enterotoxin B (SEB) in foods was decreased by food constituents that react with rabbit anti-SEB with an equivalency of over 2 ng/ml. This activity was minimized by a conditioning step for anti-SEB and by removal of interfering compounds in the sample by extraction. The assay was a sequential solid-phase radioimmunoassay technique in which polystyrene test tubes were initially incubated with antisera and then with bovine serum albumin. The tubes were then conditioned with either a centrifuged aqueous cheese extract or, equally effective, reconstituted nonfat dry milk for 16 h at 4 degrees C. Samples of milk or heat-treated and CHCl(3)-extracted cheese or chicken salad slurries were incubated in the assay tubes for 6 h at 37 degrees C. The samples were replaced by I-labeled SEB and incubated for a further 2 to 4 h before the contents were removed and the tubes were washed and counted. A buffer solution containing known concentrations of toxin served as standards for assaying SEB in the food extracts. The entire assay can be accomplished within 24 h with a sensitivity of 1 ng/ml in milk and in the cheese extract or 1.3 ng/ml in the chicken salad extract.