Rotaviruses collected in Bangladesh during 1985 to 1986 were culture adapted and used in a comparative serotyping study with three groups of monoclonal antibodies, all of which reacted with the major neutralization protein (VP7) of serotype 1, 2, 3, or 4. The goals were to determine which monoclonal antibodies most accurately predicted the serotype and why large variations in serotyping efficiencies have occurred with these monoclonal antibodies in previous studies. The 143 rotavirus isolates used in this study belonged to 69 different electropherotypes; and 44, 23, 21, and 55 isolates were identified as serotype 1 through 4, respectively, by neutralization with serotype-specific hyperimmune antisera. Serotyping specificity by enzyme-linked immunosorbent assay with monoclonal antibodies was 100% consistent with results found by neutralization with polyclonal antisera, but large differences were observed in the sensitivities of the different monoclonal antibodies. Monoclonal antibodies 5E8 (serotype 1), 1C10 (serotype 2), 159 (serotype 3), RV3:1 (serotype 3), ST-3:1 (serotype 4), and ST-2G7 (serotype 4) reacted with all the isolates of the corresponding serotype for which there were sufficient infectious particles. Monoclonal antibody 2F1 (serotype 2) was much less sensitive and reacted with only five serotype 2 isolates, but these were among those with the highest titers. Monoclonal antibodies RV4:2 (serotype 1), KU6BG (serotype 1), RV5:3 (serotype 2), and S2-2G10 (serotype 2), on the other hand, failed to react with between one and three isolates of the corresponding serotypes which had high titers, apparently because of epitope changes in these isolates. Effects of epitope variation were, however, most apparent with monoclonal antibodies 2C9 (serotype 1) and YO-1E2 (serotype 3), which reacted with one and no isolates of the corresponding serotypes, respectively. Cross-neutralization of escape mutants indicated that the serotype 1 monoclonal antibodies 5E8, 2C9, and RV4:2 reacted with different but probably overlapping epitopes, as did serotype 2 monoclonal antibodies 2F1, 1C10, and RV5:3, finding that were consistent with the enzyme-linked immunosorbent assay data. Because of epitope variations between rotavirus strains, serotyping with several monoclonal antibodies directed at different epitopes may increase the sensitivity of the method.