Laboratory-attenuated strains of each of the four dengue serotypes previously tested as monovalent vaccines in volunteers were combined and tested for immunogenicity, safety, and reactogenicity in 16 dosage combinations. Tetravalent vaccines made using combinations of high (10(5-6) plaque-forming units [PFU]/dose) or low (10(3.5-4.5) PFU/dose) dosage formulations of each of the four viruses were inoculated in 64 flavivirus non-immune adult volunteers to determine which, if any, formulation raised neutralizing antibodies in at least 75% of volunteers to at least three of four dengue serotypes following one or two inoculations. Such formulations, if safe and sufficiently non-reactogenic, would be considered for an expanded Phase II trial in the future. Formulations 1-15 were each inoculated into three or four volunteers (total = 54) on days 0 and 28. Formulation 16 was tested in 10 volunteers, five volunteers inoculated on days 0 and 30, one volunteer on days 0 and 120, and four volunteers on days 0, 30, and 120. Blood was drawn for serologic assays immediately before and one month after each vaccination, and for viremia assay on day 10 after each vaccination. The 16 formulations were safe, but variably reactogenic after the first vaccination, and nearly non-reactogenic after the second and third vaccinations. Reactogenicity was positively correlated with immunogenicity. Similar proportions of volunteers seroconverted to dengue-1 (69%), dengue-2 (78%), and dengue-3 (69%), but significantly fewer volunteers seroconverted to dengue-4 (38%). The geometric mean 50% plaque reduction neutralization test titers in persons who seroconverted were significantly higher to dengue-1 (1:94) than to dengue-2 (1:15), dengue-3 (1:10), and dengue-4 (1:2). Seven formulations met the serologic criteria required for an expanded trial, and three of these were sufficiently attenuated clinically to justify further testing.