As the sensitivity of the conventional techniques for identifying Shigella spp. and enteroinvasive Escherichia coli (EIEC) causing dysentery cases is low, a PCR assay was evaluated in this study. Analytical sensitivity (2 x 10(2) cfu) of the PCR technique was obtained by artificially spiking negative stool samples with a standard strain of S. flexneri type 2, then determining the detection limit. Specificity (100%) of the method was determined by testing a number of known Shigella and EIEC strains and organisms other than Shigella spp. A total of 300 stool samples collected from children with acute diarrhoea was plated on to two selective agar media after enrichment in Luria broth. Shigella spp. were isolated from 7.7% (23 of 300) and EIEC from 1% (3 of 300) patients. All enriched stool samples were subjected to PCR to amplify the target sequence of invasive plasmid antigen (ipa)H locus, a multicopy element found on the chromosome and invasion plasmid. The stool PCR was positive in 24 of the 26 culture-positive and in 22 culture-negative stools, thus detecting the presence of Shigella spp. or EIEC in 15.3% (46 of 300) of diarrhoea cases. When an ial probe was used for colony hybridistion with enriched stool cultures blotted on to membranes, 9.6% (29 of 300) of dysentery cases were identified as being caused by Shigella spp. or EIEC. Thus the sensitivity of enriched stool culture, colony hybridisation and enriched stool PCR was found to be 54%, 60% and 96%, respectively, when each of the methods was compared to the total microbiologically confirmed cases of dysentery. It was also observed that only 38% (48 of 126) of acute bloody dysentery cases actually had shigella or EIEC infection, as confirmed by laboratory methods. Moreover, this PCR assay could identify a number of untypable Shigella strains (Sh OUT), which would have remained undiagnosed had this assay not been used.