Pure-culture broth-based models of the growth of Escherichia coli O157:H7 have been used to estimate its behavior in ground beef, even though these models have not been adequately validated for this food product. This situation limits accurate estimates of the behavior of E. coli O157:H7 in ground beef and introduces uncertainties in risk assessments. In the present study, the growth of single and multiple strains of E. coli O157:H7 were measured in retail ground beef stored at 10 degrees C for up to 12 days, and the results were compared with estimates generated by the U.S. Department of Agriculture's Pathogen Modeling Program (PMP; version 5.1). At pH 5.9, the PMP predicted a maximum population density (MPD) of 9.13 log10 CFU/g, an exponential growth rate (EGR) of 0.052 log10 CFU/h, and a lag time of 56.3 h. Similar parameter values were observed for sterilized ground beef; however, no lag phase was observed. In contrast, the mean MPD and EGR for retail ground beef were 5.09 log10 CFU/g and 0.019 log10 CFU/h, respectively, and no lag phase was observed. Both the EGR and the MPD increased with decreasing fat levels. There was low variation in the MPD and EGR parameters for the nine E. coli O157:H7 ground beef isolates. Two isolates of competitive native flora were separately added to sterilized ground beef, and the EGR and MPD decreased as the ratio of competitive flora to E. coli O157:H7 increased. For one strain, at ratios of 1:1, 10:1, and 100:1, the EGRs were 0.033, 0.025, and 0.018 log10 CFU/h, respectively, and the MPDs were 6.14, 5.08, and 4.84 log10 CFU/g, respectively. These results demonstrate that existing broth-based models for E coli O157:H7 must be validated for food and that models should consider the effects of the food matrix, the competitive microflora, and potential pathogen strain variation.