Numbers of Escherichia coli O157 in food may be low and sensitive techniques are therefore needed for its detection. The objectives of this study were to use carcass meat samples artificially inoculated with various strains of E. coli O157 to compare the sensitivity of enrichment in three different media and to compare immunomagnetic separation followed by culture of magnetic beads to cefixime tellurite sorbitol MacConkey agar with three immunoassays for the detection of E. coli O157 in the enrichment cultures. Duplicate 250, 25 and 2-3 CFU of each of 16 strains of E. coli O157 added to 25-g samples of beef carcass meat were used to compare the sensitivity of (1) enrichment in supplemented tryptone soya broth (sTSB), Reveal 8-h and Reveal 20-h media, and (2) immunomagnetic separation and culture to cefixime tellurite sorbitol MacConkey agar (IMS/CT-SMAC) with Reveal, VIP and STAT immunoassays for detecting the organism. An initial inoculum of 250 CFU/25 g meat was detected in all 32 samples by IMS/CT-SMAC performed on all enrichment media and by Reveal performed on Reveal 8-h and Reveal 20-h media, but in only 30, 19 and 9 samples by Reveal, VIP and STAT, respectively, performed on sTSB medium. An initial inoculum of 25 CFU/25 g meat was detected in 28, 32 and 30 of 32 samples by IMS/CT-SMAC performed on sTSB, Reveal 8-h and Reveal 20-h media, respectively, and in 32 and 30 samples by Reveal performed on Reveal 8-h and Reveal 20-h media, but in only 22, 11 and 2 samples by Reveal, VIP and STAT, respectively, performed on sTSB medium. An initial inoculum of 2-3 CFU/25 g meat was detected in 25, 31 and 28 of 32 samples by IMS/CT-SMAC performed on sTSB, Reveal 8-h and Reveal 20-h media, respectively, and in 25 and 23 samples by Reveal performed on Reveal 8-h and Reveal 20-h media, but in only 14, 1 and 0 samples by Reveal, VIP and STAT, respectively, performed on sTSB medium.