In an attempt to develop a standard for ELISA-PCR detection of Shiga toxin producing Escherichia coli (STEC) O157, six published PCR tests were tested in a comparative study on a panel of 277 bacterial strains isolated from foods, animals and humans. These tests were based on the detection of the genes rfbE [J. Clin. Microbiol. 36 (1998) 1801] and rfbB [Appl. Environ. Microbiol. 65 (1999) 2954], the 3' end of the eae gene [Epidemiol. Infect. 112 (1994) 449], the region immediately flanking the 5' end of the eae gene [Int. J. Food. Microbiol. 32 (1996) 103], the flicH7 gene [J. Clin. Microbiol. 35 (1997) 656], or a part of the recently described 2634-bp Small Inserted Locus (SILO(157) locus) of STEC O157 [J. Appl. Microbiol. 93 (2002) 250]. Unlike the other PCR assays, those amplifying the rfb sequences were unable to distinguish toxigenic from nontoxigenic O157. These assays were relatively specific to STEC O157, giving essentially a cross reaction with clonally related E. coli O55 and to a lesser extent with E. coli O145, O125, O126. They also detected the Shiga toxin (stx)-negative derivatives of STEC O157. Based on these results, an ELISA-PCR assay consisting of the solution hybridization of amplicons with two probes that ensured the specificity of the amplification was developed. The ELISA-PCR assay, which used an internal control (IC) of inhibition, was able to detect 1 to 10 copies of STEC O157 in the PCR tube. Adaptation of PCR into ELISA-PCR assay format facilitates specific and sensitive detection of PCR amplification products and constitutes a method of choice for screening STEC O157.