The gastrointestinal tract is considered to be a major route of infection for human immunodeficiency virus (HIV). Infection of human colon epithelial cells by HIV is not blocked by anti-CD4 antibodies known to block infection of lymphoid cells (J. Fantini, N. Yahi, and J. C. Chermann, Proc. Natl. Acad. Sci. USA 88:9297-9301, 1991), suggesting the presence of an alternate receptor for HIV on these cells. In this report, we show that (i) a monoclonal antibody specifically directed against galactosyl ceramide inhibited the infection of HT29 cells by two markedly different strains of HIV-1, as assessed by polymerase chain reaction amplification and reverse transcriptase assay; (ii) this antibody strongly labeled the surface of HT29 cells by immunofluorescence and electron microscopic immunolocalization; (iii) the labeling was preferentially but not totally restricted to the basolateral membrane domain of differentiated colonic cells, in agreement with the ability of HIV to infect both the apical and basolateral surfaces of these epithelial cells; and (iv) in thin-layer chromatography-immunostaining experiments with neutral glycolipids prepared from HT29 cells, the antibody specifically reacted with a ceramide monoglycoside fraction corresponding to galactosyl ceramide. We did not detect this glycolipid in lymphoid cells, and anti-galactosyl ceramide antibodies consistently failed to inhibit HIV infection of these cells. These data suggest that galactosyl ceramide (or a derivative) is an essential component of the receptor for HIV on the surface of HT29 cells.