Among 315 fecal specimens collected from children hospitalized with diarrhea in Chiang Mai, Thailand, in 2000-2001, group A rotavirus was detected in 107 (34.0%). Of these, 98 (91.6%) were G9, 6 (5.6%) were G3 and 3 (2.8%) were G2, respectively. Identification of their P-types demonstrated that 103 (96.3%) were P[8], 3 (2.8%) were P[4], and 1 (0.9%) was P[3] genotypes. Determination of G- and P-type combination revealed that all of G9 isolates were associated with P[8]. G9P[8] was the most predominant genotype and accounted for the majority (91.6%) of rotaviruses detected in this study. Molecular characterization of these G9 isolates demonstrated that all had long electropherotype, 96 of 98 (98.0%) belonged to subgroup II, one belonged to subgroup I and the other one was subgroup unidentifiable. All of G9 isolates possessed NSP4 genetic group B except for one isolate that showed dual genetic group specificities, B and C. The full-length VP7 gene nucleotide sequences among 15 representatives of these G9 strains were found to be highly homologous with percent identities of 99.3%-100%. Comparison with other G9 strains recently isolated showed that their nucleotide sequences were closely related to those of the US strain, US1205 (98.7%-99.0%) and Thai strain, 97CM108 (98.1%-99.0%). Interestingly, they were most closely related to the Japanese strain, 00-SG2509VP7, isolated in the same epidemic season, with percent nucleotide sequence identity of 99.4%-99.8%. The data imply that G9 strains isolated in this study and a G9 strain isolated in Japan in the year 2000 might have descended from the same ancestor.