Several reports implicate Langerhans cells of skin as susceptible targets, reservoirs, and vectors for transmission of HIV: 1) numbers of Langerhans cells in skin of HIV-infected patients were decreased about 50% of that in control skin; 2) as many as 30% of Langerhans cells in the skin of HIV-infected patients were morphologically abnormal; 3) viral particles typical for HIV were identified in or around 2 to 5% of these cells; and 4) infectious HIV was isolated from skin biopsies of infected patients. These results were consistent with similar observations of HIV-infected macrophages in such tissues as brain, lung, and lymph node. Despite these findings, other investigators find no evidence for virus infection in the epidermis of HIV-infected patients by any of several immunohistochemical or ultrastructural criteria. To address this controversy, we obtained skin from 28 HIV-seropositive subjects at various clinical stages by full thickness biopsy or suction blister. Samples were analyzed by transmission electron microscopy for presence of HIV virions, by immunofluorescent staining for viral proteins, by in situ hybridization for HIV-specific mRNA, by polymerase chain reaction amplification of virus-specific DNA, and by direct virus isolation by coculture of epidermis onto monocyte target cells. By any of these techniques, demonstration of HIV in the epidermis of infected patients was equivocal and even then, infrequent. In contrast, viral DNA was detected from the dermis of the same skin samples (26 of 28 samples). Moreover, the number and morphology of Langerhans cells in skin of infected patients were within normal limits, regardless of stage of disease. These studies in toto suggest that a role for Langerhans cells as a principal viral reservoir or vector of transmission is highly unlikely.