Water samples were concentrated by the modified adsorption-elution technique followed by speedVac reconcentration of the filter eluates. Reverse transcriptase-nested polymerase chain reaction (RT-nested PCR) was used to detect rotavirus RNA in concentrates of the water. The detection limit of the rotavirus determined by RT-nested PCR alone was about 1.67 plaque forming units (PFU) per RT-PCR assay and that by RT-nested PCR combined with concentration from 1l seeded tap water sample was 1.46 plaque forming units per assay. Water samples were collected from various sources, concentrated, and determined rotavirus RNA. Of 120 water samples, rotavirus RNA was detected in 20 samples (16.7%); 2/10 (20%) of the river samples, 8/30 (26.7%) of the canal samples, and 10/40 (25%) of the sewage samples but was not found in any tap water samples (0/40). Only three water samples were positive for rotavirus antigen determined using an enzyme-linked immunosorbent assay (ELISA). Alignment analysis of the sequenced PCR product (346-bp fragment) was performed in eight rotavirus-positive samples using the rotavirus sequence deposited in the GenBank. All samples gave the correct VP7 sequence. Results of analysis showed two samples similar to human rotavirus (97-98%), five similar to rotavirus G9 sequence (94-99%), and one sample similar to animal rotavirus (97%). PCR inhibitors were not observed in any concentrated water samples. In all 20 (of 120) samples where rotaviruses were found, fecal coliforms including Escherichia coli were also found, but of the samples testing negative for rotaviruses, 76 were fecal coliforms positive and 69 were E. coli positive. The combination of the virus concentration method and RT-nested PCR described below made it possible to effectively detect rotaviruses in environmental water samples.