Fecal Escherichia coli isolates from 196 patients with watery diarrhea and 68 healthy individuals (controls) were analyzed in Bangladesh immediately after isolation for the presence of colonization factor antigen (CFA) I or II (CFA/I or CFA/II, respectively) by a mannose-resistant hemagglutination (MRHA) test with six species of erythrocytes and by a slide agglutination test with absorbed CFA/I or CFA/II antisera. The presence of CFAs was confirmed by immunodiffusion analyses done in Sweden. By these methods, it was found that 49 of 69 enterotoxin-producing E. coli strains isolated from patients carried CFA/I or CFA/II, whereas none of the nonenterotoxigenic E. coli isolates or the three toxin-positive strains isolated from healthy individuals carried these adhesins. All E. coli strains retained their MRHA ability after transportation to Sweden followed by one subculture and after storage at -70 degrees C (but not at room temperature) for 1 to 2 years without further subculturing. After 5 to 10 subcultures of the fresh isolates, however, 70% of the initially CFA/I- and 80% of the initially CFA/II-carrying strains analyzed did not hemagglutinate. The efficacy of different methods for detecting CFAs on the fresh isolates was compared with that of immunodiffusion. The sensitivity of MRHA with human blood group A erythrocytes for the detection of CFA/I was high (97%), but the specificity was only 69%. The sensitivity of MRHA with bovine erythrocytes for the detection of CFA/II in Bangladesh was very low but increased considerably when chicken erythrocytes were also used. Whereas both false-positive and false-negative reactions were obtained when absorbed CFA antisera were used for agglutination, antisera against purified CFAs were equally effective as immunodiffusion in identifying CFA/I and CFA/II-carrying strains.