A 3-year clinical, virologic, and serologic study of 639 children admitted to Children's Hospital and the SEATO Clinical Research Center, Bangkok, was undertaken to define the entity “Thai hemorrhagic fever” as diagnosed by house staff. Two groups of control patients were studied to measure “background” dengue and chikungunya infections in the hospital population. Of 628 patients with hemorrhagic fever with specimens adequate for study, 32% were in shock, 83 and 7.6% had dengue and chikungunya illnesses, respectively, and no etiology could be established in 12%. In each year, more than 80% of patients with hemorrhagic fever had confirmed dengue infections, but only 19% of febrile patients admitted for diagnoses other than hemorrhagic fever had dengue, and 4% may have contracted dengue in hospital. More than 85% of patients with dengue had a secondary type of antibody response to other group B arboviruses. In chikungunya disease, the onset of symptoms was more abrupt, the febrile course shorter, and maculopapular rashes, conjunctival injection, and arthralgia were more common than in dengue; the frequency of a positive tourniquet test, scattered petechiae, and epistaxis was similar, but shock and gastrointestinal hemorrhage occurred only in dengue patients. We concluded that Thai hemorrhagic fever had a varied etiology and diverse clinical manifestations. A redefinition of “hemorrhagic fever” to include only shock cases excluded chikungunya cases. When the illness was characterized by fever and mild hemorrhagic manifestations, it was impossible to distinguish in the individual case among dengue, chikungunya, and other infections.