A method to validate enumeration of Escherichia coli O157 in fecal samples from feedlot cattle was developed in these studies. Due to background flora, bovine fecal sample enumeration cannot be performed by simple direct plating techniques. Known quantities of E. coli O157:H7 were inoculated into feces, and populations were determined by direct plating of the cocktail (studies 1, 2, and 3) and manure and cocktail (studies 4 and 5) mixtures and compared with a most-probable-number (MPN)-immunomagnetic separation (IMS) method. The three-tube MPN combined preenrichment in gram-negative broth with confirmation using IMS. Five separate enumeration studies (study 1, sterile feces inoculated with 10(2) E. coli O157:H7 per g; study 2, nonsterile feces inoculated with 10(3) E. coli O157:H7 per g; study 3, nonsterile feces inoculated with 10(1) E. coli O157:H7 per g; study 4, sterile feces inoculated with 10(4) streptomycin-resistant E. coli O157:H7 per g; and study 5, sterile feces inoculated with 10(2) streptomycin-resistant E. coli O157:H7 per g) were conducted. These studies were performed to determine the precision, accuracy, and specificity at low and high levels of pathogen contamination in feces, using direct plating compared with the MPN-IMS methodology tested. There was an overall difference (P < 0.01) between direct plating and MPN-IMS methodologies, but this difference was biologically negligible due to the difference in least-squares means (0.29 +/- 0.10) being so low. The direct plating and MPN-IMS methods were correlated (r = 0.93). These results suggest that using the MPN-IMS procedures is an effective method of estimating E. coli O157 populations in naturally infected bovine fecal samples.