We compared the usefulness of gene probes with standard bioassays to identify diarrhoeagenic Escherichia coli amongst isolates from Bangladeshi children under 1 year of age with diarrhoea. E. coli isolates were analysed with specific gene probes for localised adhesiveness (LA), diffuse adhesiveness (DA), heat-labile toxin (LT), heat-stable toxin (ST), Shiga-like toxins (SLT I and SLT II), and enteroinvasiveness, and in bioassays for production of enterotoxins and cytotoxins, and for cell adherence. With 1136 isolates from 387 patients, there was general agreement between the two assay methods. When there was disparity, gene-probe-positive isolates gave negative results in the corresponding bioassay. In the HeLa cell adherence assay, 94% of the LA probe-positive isolates and 91.6% of the DA probe-positive isolates gave positive bioassay results for LA and DA respectively. Thirty-six of 39 LT probe-positive isolates and 73 of 86 ST probe-positive isolates gave positive results in the bioassays. Of 28 isolates that gave negative results in the suckling mouse assay but were initially positive with the probe for ST, 15 were later found to hybridize with the cloning vector for the ST probe. Addition of denatured vector DNA at a concentration of 10 micrograms/ml in the hybridisation solution eliminated these false positive results. None of the other probe-positive isolates hybridised with any of the cloning vectors used. The DNA hybridisation assay appeared to be a convenient alternative to bioassays for screening large numbers of isolates in epidemiological investigation.