A total of 1,385 stool specimens were collected from children with diarrhea at two hospitals in Wuhan, Hubei Province, China, in 1994 and 1995, and screened for rotavirus by polyacrylamide gel electrophoresis of viral RNA. Group A rotavirus was detected with high frequency; 56.5% (87/154) and 40.8% (502/1,231) of the specimens collected in 1994 and 1995, respectively, were positive for rotavirus. Assignment of G serotype and P type (VP4 genotype) of group A rotavirus by ELISA with monoclonal antibodies and/or PCR, respectively, showed that strains of G2-P[4] and G1-P[8] specificity were predominant in 1994 and in 1995, respectively. In contrast, a single strain was found to have a P[9] type specificity, and no G4 strain was detected. Unusual combinations of RNA pattern-subgroup-G serotype-P type, such as long pattern-subgroup I-G1-P[8], short pattern-subgroup II-G3-P[4] and short pattern-subgroup I-G1-P[4], were detected in four specimens. Nucleotide sequences of the VP8* and/or NSP5 genes from two Chinese P[8] strains 470 and 582 and one Chinese P[9] strain 512 as well as five Japanese P[9] strains (K8, AU1, M318, 0264, and 0265) were determined and compared with the published sequences of the corresponding gene. In the phylogenetic tree of VP8* sequences of P[9] strains, which formed two clusters each having strain K8 or AU-1 as the representative strain, the Chinese P[9] strain was found in the cluster represented by AU-1, although it was most distantly related to other strains. While NSP5 sequences of human strains with P[9] specificity were related to simian and bovine strains, that of Chinese P[8] strains was most closely related to those of porcine strains. A single group C rotavirus (No. 208) was detected. Nucleotide sequences of its VP4, VP6, VP7, and NSP4 genes were very similar to those of group C human rotaviruses detected worldwide.