Rotaviruses were prospectively studied in 51 rural Costa Rican children from birth to two years. Samples of feces were collected weekly over a 33-month period. Rotavirus was detected in 45 (1.04%) of 4,317 fecal specimens; 39 infections were documented (an incidence of 0.5 infection per child-year), only five of which were associated with diarrhea (a pathogenicity of 12.8%). Secretory antibody in fecal extracts, detected in six of 39 infections, was short lived and did not protect against reinfection. Serum antibody was present in 69.6% of two-year-old children, but was not detected in 18.8% with documented infections. On the other hand, serum antibody was present in six of 14 children in whom rotavirus was not detected, thus increasing the overall incidence to 0.6 infection per child-year. The combination of prolonged breast-feeding, exposure to a lower infecting dose (compared with urban children), and a higher standard of hygiene than expected may explain the low incidence and low pathogenicity of rotavirus among these rural children.The epidemiology of rotavirus infection in populations observed in their own environment needs further elucidation, particularly in less-developed countries where diarrheal disease takes its highest toll. In Costa Rica, rotavirus was demonstrated as the most important etiologic agent of acute diarrhea among outpatients at the National Children's Hospital in San Jose [1]. Hospital-based studies are of limited epidemiological value, however, because they are cross-sectional in nature; cover, by definition, the most serious cases that require hospital treatment; and thus, do not reflect the situation in the general population. Consequently, a cohort study was designed in order to study patterns of infection and immunity to rotavirus under natural conditions. This longitudinal study permitted investigation of the