The efficiency of the pooling method for the detection of enterotoxigenic Escherichia coli (ETEC) strains associated with children acute diarrhea, was evaluated. This study involved the analysis of 6989 E. coli strains corresponding to 1485 cases, coming from 7 hospitals at different geographic locations. Three to five strains from each case were inoculated in pool in Casamino-acids-yeast extract-salts medium plus lincomycin (30 micrograms/ml) and incubated at 37 degrees C during 18 hours with shaking. Polymyxin sulphate B (2200 U/ml) was added to the cultures, and incubation with shaking continued for additional 30 minutes. The culture was then centrifuged and the supernatant tested by a) the enzyme-linked immunosorbent assay to detect heat-labile enterotoxin (LT) and b) the suckling mouse assay for the detection of heat-stable enterotoxin (ST). Strains from a pool were individually studied in all positive cases as shown by the pooling methodology. Also, in 1 out of every 15 negative cases by the pool method, their component strains were individually analyzed, to confirm that there were no false negatives. Fifty seven LT-ETEC, 61 ST-ETEC and 15 LT-ST cases were detected. From the 89 negatives selected by the pool method, 2 cases were positive when the component strains were individually tested (p1 = 2.25%, 95% confidence interval from 0.27 to 7.88%). In both cases one of the non ETEC strains inhibited the ETEC strain. In other 5 negative cases (p2 = 5.62%, 95% confidence interval from 1.85% to 12.63%), ETEC strains were detected when analyzed individually, and gave a positive result when the pooling method was repeated.(ABSTRACT TRUNCATED AT 250 WORDS)