Dengue infections were prospectively studied among 4- to 16-year-old students at a Bangkok school. Blood samples were obtained from 1,757 students in June 1980, before the dengue season, and in January 1981, after the season, and tested for dengue antibodies by the hemagglutination inhibition method. Classrooms were monitored daily for school absences. Fifty percent of the children had antibodies to, and were presumably immune to, at least 1 dengue serotype by the age of 7 years. Most (90/103, 87%) students who became infected by dengue viruses during the study period were either asymptomatic or minimally symptomatic (absent only 1 day). Most (7/13, 53%) of the symptomatic dengue infections (absent with fever for ≥2 days) were clinically recognized as cases of dengue hemorrhagic fever which required hospitalization. None of 47 primary dengue infections required hospitalization, whereas 7 of 56 secondary infections did (P = 0.012). Preexistent dengue immunity, as detected by conventional serologic techniques, was a significant (odds ratio ≥ 6.5) risk factor for development of dengue hemorrhagic fever.