A regional prospective study of the epidemiology of bluetongue virus (BTV) serotypes covering 11 countries in Central America and the Caribbean took place between 1987 and 1992. Active surveillance revealed BTV infection to be endemic in the absence of confirmed indigenous cases of bluetongue. During the 6-year span of the study, over 300 BTV isolations were obtained from cattle and sheep. Results of the earlier years of the study were summarized, and surveillance activities in the concluding months of the study from November 1990 to February 1992 were evaluated. Forty-five BTV isolations were made during this time, 44 from sentinel cattle and 1 from a ram with clinical signs compatible with contagious ecthyma. Virus isolation from potential vectors also was attempted, yielding a further 9 BTV isolates from parous Culicoides insignis and C pusillus, 2 BTV isolates from blood-engorged C filarifer, and 1 epizootic hemorrhagic disease virus type-2 isolate from parous C pusillus. Our extensive network of sentinel herds in the region detected BTV-1 as the predominant serotype in Central America in 1991, after an apparent absence of 1 year in the sentinel animals. Other serotypes in Central America at that time included BTV-3 and BTV-6. In Puerto Rico and the Dominican Republic, BTV-4 became the predominant serotype, without detection of BTV-8 and BTV-17, which were common in recent years of the study. The serotypes found in the Caribbean Basin continued to have marked differences from those in North America. The importance of viewing bluetongue as an infection, the distribution of which is determined principally by ecologic factors, is emphasized.