Samples of commercially pelleted poultry feed (30 g) were inoculated with nalidixic acid-resistant marker strains of Salmonella heidelberg or S. montevideo at levels of 1 (low), 20 (medium), or 40 (high) cells/g of feed, then 100 ml of either a nonselective preenrichment medium (lactose broth) or a selective enrichment medium [selenite cystine (SC) or tetrathionate brilliant green (TT) broth] was added and incubated at either 37 or 43 C for 24 or 48 hr in different experiments. Four selective plating media [MacConkey with 100 ppm nalidixic acid, brilliant green (BG) sulfa, modified BG, bismuth sulfite] were then streaked. MacConkey agar with nalidixic acid served as a control to which the other three plating media were compared. After 24 hr incubation at 37 C, colonies with characteristics typical of Salmonella were selected from the plates and examined biochemically and serologically to determine if they were the marker organism. In this study, preenrichment was counterproductive, because the marker organisms were recovered much more often with direct enrichment in SC or TT broth than with reenrichment in lactose broth. The TT broth produced 135 positive recoveries, whereas only 88 positive recoveries were made with SC. No differences were observed between 37 and 43 C or 24 and 48 hr incubation of the selective enrichment media. Overall efficiency of the plating media for Salmonella recovery from feed, when compared to the control (MacConkey), was BG sulfa, 65%;modified BG, 64%; and bismuth sulfite, 40%. The most effective enrichment broth-plating medium combination was TT-BG sulfa, yielding 86% positive recoveries when compared to the control.