Abstract PCR was compared to DNA hybridization, ELISAs and bioassays for the identification of enterotoxigenic Escherichia coli. Seventy E. coli strains of human and porcine origine were examined for heat-stable toxin (STI) and heat-labile toxin (LT). Hybridization was performed with two commercial alkaline phosphatase-conjugated oligonucleotide probes and two longer PCR-amplified radioactive probes. The adrenal cell test and the infant mouse test were used as reference methods. The LT PCR and hybridization using the radioactive LT probe detected all strains of both human and porcine origin, while the LT specific oligoprobe and the LT ELISA detected 62% and 59% of the strains, respectively. The agreement of the results of the different STI assays was generally higher for human than porcine isolates. The radioactive probe and the ELISA, with detection levels of 100% and 95%, respectively, correlated best to the bioassay for the detection of the humans STI strains. PCR and the STI specific oligoprobe however, detected only 80% and 75%, respectively. The corresponding figures for the porcine STI isolates were considerably lower. A great variation was demonstrated among the genes for STI among the isolates of both origin. It was concluded that the PCR was a fast and reliable method for the identification of LT. The STI PCR did not, however, fulfil the desired criteria, probably due to the variability in the sTI genes. It was also concluded that DNA hybridization using probes of at least 322 bases for LT and 175 bases for STI is useful for the verification of cultivated samples.