Complement fixation (CF), indirect double-antibody sandwich (DAS) enzyme-linked immunosorbent assay (ELISA) and solid-phase immune electron microscopy (SPIEM) were compared for their ability to subgroup 73 human rotavirus (HRV) strains from infants and young children with gastroenteritis admitted to one or the other of two different hospitals of Northern Italy. By both indirect DAS ELISA and SPIEM all 73 HRV strains were classified into one or the other of two subgroups. By CF only 67 strains could be subgrouped, as six HRV-positive stools showed anticomplementary activity which could not be eliminated. Indirect DAS ELISA required subgroup-specific, unabsorbed antisera from two different animal species. For SPIEM two antisera from a single animal species were needed, but they had to be absorbed with single-shelled bovine rotavirus for HRV subgrouping to be reliable. Indirect DAS ELISA appeared to be the technique most suitable for extensive application in epidemiological studies of HRV infections by different subgroups. However, SPIEM allowed rapid subgrouping of HRV in stool specimens showing anticomplementary activity in the CF test or non-specific reactions in the ELISA test. In one area of Northern Italy the prevalence of subgroup I HRV infections was 7.8 per cent, while in another it reached 68.1 per cent in the same period.