Suspensions of 24 rotavirus strains, 6 for each known human rotavirus serotype, were serially diluted and titrated by (i) enzyme-linked immunosorbent assay (ELISA) for rotavirus detection, using monoclonal antibodies (MAbs) specific for group-specific sites of the VP6 inner capsid protein; (ii) ELISA for subgrouping, using MAbs reactive with subgroup-specific determinants of rotavirus VP6; (iii) ELISA for serotyping, using MAbs directed to serotype-specific sites of the VP7 outer capsid glycoprotein; and (iv) solid-phase immune electron microscopy (SPIEM) for serotyping, using VP7-specific MAbs. In addition, in each preparation the proportion of double-shelled rotavirus particles were determined by direct electron microscopy. Results showed that SPIEM was 2- to 16-fold more sensitive than ELISA for serotyping of rotavirus. The titers in VP7-specific tests correlated well with the proportion of double-shelled virus particles in each of the samples. Titers obtained by ELISA for serotyping of suspensions containing 20% or fewer complete particles were up to 4,096-fold lower than those obtained by ELISA for detection. ELISA serotyping titers of samples containing 20 to 80% double-shelled rotavirus particles were up to 128-fold lower than ELISA detection titers, whereas preparations with nearly 100% complete particles had ELISA titers that were less different from each other. ELISA subgrouping titers were four- to eightfold lower than corresponding rotavirus detection titers. It was concluded that, although SPIEM appears to be more sensitive than ELISA, the amount of complete virus particles in the specimens is of critical importance for successful serotyping of human rotavirus strains. Samples rich in single-shelled particles but containing low amounts of VP7 outer capsid glycoprotein might even be strongly reactive in assays for rotavirus detection and subgrouping but virtually unreactive in tests for serotyping.