We report the first isolation and molecular and antigenic characterization of a human ehrlichial species in South America. A retrospective study was performed with serum specimens from 6 children with clinical signs suggestive of human ehrlichiosis and 43 apparently healthy adults who had a close contact with dogs exhibiting clinical signs compatible with canine ehrlichiosis. The evaluation was performed by the indirect fluorescent-antibody assay with Ehrlichia chaffeensis Arkansas, Ehrlichia canis Oklahoma, and Ehrlichia muris antigens. The sera from two apparently healthy humans were positive by the indirect fluorescent-antibody assay for all three antigens. Of the three antigens, samples from humans 1 and 2 showed the highest antibody titers against E. chaffeensis and E. muris, respectively. The remaining serum samples were negative for all three antigens. One year later examination of a blood sample from subject 1 revealed morulae morphologically resembling either E. canis, E. chaffeensis, or E. muris in monocytes in the blood smear. The microorganism, referred to here as Venezuelan human ehrlichia (VHE), was isolated from the blood of this person at 4 days after coculturing isolated blood leukocytes with a dog macrophage cell line (DH82). The organism was also isolated from mice 10 days after intraperitoneal inoculation of blood leukocytes from subject 1. Analysis by electron microscopy showed that the human isolate was ultrastructurally similar to E. canis, E. chaffeensis, and E. muris. When the virulence of VHE in mice was compared with those of E. chaffeensis, E. canis, and E. muris, only VHE and E. muris induced clinical signs in BALB/c mice at 4 and 10 days, respectively, after intraperitoneal inoculation. VHE was reisolated from peritoneal exudate cells of the mice. Only E. chaffeensis- and E. muris-infected mice developed significant splenomegaly. Western immunoblot analysis showed that serum from subject 1 reacted with major proteins of the VHE antigen of 110, 80, 76, 58, 43, 35, and 34 kDa. Human serum against E. chaffeensis reacted strongly with 58-, 54-, 52-, and 40-kDa proteins of the VHE antigen. Anti-E. canis dog serum reacted strongly with 26- and 24-kDa proteins of VHE. In contrast, anti-E. sennetsu rabbit and anti-E. muris mouse sera did not react with the VHE antigen. Serum from subject 1 reacted with major proteins of 90, 64, or 47 kDa of the E. chaffeensis, E. canis, and E. muris antigens. This reaction pattern suggests that this serum sample was similar to serum samples from E. chaffeensis-infected human patients in Oklahoma. The base sequence of the 16S rRNA gene of VHE was most closely related to that of E. canis Oklahoma. On the basis of these observations, we suggest that VHE is a new strain or a subspecies of E. canis which may cause asymptomatic persistent infection in humans.