Galactosyltransferase immunoreactive sites were localized in human duodenal enterocytes by the protein A-gold technique on thin sections from low temperature Lowicryl K4M embedded biopsy specimens. Antigenic sites detected with affinity-purified, monospecific antibodies were found at the plasma membrane of absorptive enterocytes with the most intense labeling appearing along the brush border membrane. The lateral plasma membrane exhibited a lower degree of labeling at the level of the junctional complexes but the membrane interdigitations were intensely labeled. The labeling intensity decreased progressively towards the basal part of the enterocytes and reached the lowest degree along the basal plasma membrane. Quantitative evaluation of the distribution of gold-particle label proved its preferential orientation to the outer surface of the plasma membrane. In addition to this membraneassociated labeling, the glycocalyx extending from the microvillus tips was heavily labeled. Occasionally, cells without plasma membrane labeling were found adjacent to positive cells. The demonstration of ecto-galactosyltransferase on membranes other than Golgi membranes precludes its general use as a marker for Golgi membrane fractions. The possible function of galactosyltransferase on a luminal plasma membrane is unclear at present, but a role in adhesion appears possible on the basolateral plasma membrane.In 1970, Roseman (24) put forward a hypothesis on a possible role of cell surface glycosyltransferases in cell adhesion and recognition. Two reviews on this subject have been published ( 18, 32) and give a comprehensive account of all experimental evidence supporting the original hypothesis. In summary, evidence for a cell surface location of glycosyltransferases is still based on indirect biochemical and autoradiographic data. Glycosyltransferase activities were found associated with plasma membranes upon fractionation (6,12,38), and orientation to the external face was assumed by their ability to glycosylate nonpermeable substrates such as glycoprotein acceptors and sugar-derivatized agarose beads (16,36,37). Autoradiographic evidence that consisted of the electron microscopic demonstration of radioactive substrates incorporated into the plasma membrane also suggested the presence of these enzymes on the cell surface but did not formally prove it (20). The conclusions based on the evidence summarized in the above cited reviews were challenged by Keenan and Morr6 (13) and Deppert et al. (8). At the present time, location ofglycosyltransferase at the plasma membrane is still an open question since no clear-cut in situ demonstration has been provided. Evidence for a putative role in adhesion and recognition, for which a conclusive demonstration of glycosyltransferases on the outer side of the plasma membrane is a prerequisite, remains circumstantial (18).Recently, monospeciflc antibodies against human galactosyltransferase (E.C. 2.4.1.22, 2.4.1.38, 2.4.1.90) became available and were used to localize this enzyme by immu...