To assess public health risks of rotavirus via drinking water consumption, a cell culture-PCR assay was developed and optimized for the detection of infectious environmental rotavirus strains in naturally contaminated source waters for drinking water production. Infectious rotavirus concentrations were estimated by an optimized cell culture-PCR assay as most probable numbers by using the presence or absence of replicated virus in different sample volumes. Infectious rotavirus was detected in 11 of 12 source water samples in concentrations varying from 0.19 (0.01-0.87) to 8.3 (1.8-34.0) infectious PCR detectable units per litre (IPDU/l), which was not significantly different from the concentrations of infectious enterovirus in these samples. In 55% of the samples, rotavirus genomes were 1000 to 10 000 times (3 log(10)-4 log(10)) more abundantly present than infectious rotavirus particles, whereas in the remaining 45% of the samples, rotavirus genomes were less than 1000 times (< 3 log(10)) more abundantly present. The broad variation observed in the ratios of rotavirus RNA and infectious particles demonstrates the importance of detecting infectious viruses instead of viral RNA for the purposes involving estimations of public health risks.