Urothelial umbrella cells are characterized by apical, rigid membrane plaques, which contain four major uroplakin proteins (UP Ia, Ib, II and III) forming UPIa/UPII and UPIb/UPIII pairs. These integral membrane proteins are thought to play an important role in maintaining the physical integrity and the permeability barrier function of the urothelium. We asked whether the four uroplakins always coexpress in the entire human lower urinary tract. We stained immunohistochemically (ABC-peroxidase method) paraffin sections of normal human ureter (n = 18) and urinary bladder (n = 10) using rabbit antibodies against UPIa, UPIb, UPII and UPIII; a recently raised mouse monoclonal antibody (MAb), AU1, and two new MAbs, AU2 and AU3, all against UPIII; and mouse MAbs against umbrella cell-associated cytokeratins CK18 and CK20. Immunoblotting showed that AU1, AU2 and AU3 antibodies all recognized the N-terminal extracellular domain of bovine UPIII. By immunohistochemistry, we found that in 15/18 cases of human ureter, but in only 2/10 cases of bladder, groups of normal-looking, CK18-positive umbrella cells lacked both UPIII and UPIb immunostaining. The UPIb/UPIII-negative cells showed either normal or reduced amounts of UPIa and UPII staining. These data were confirmed by double immunofluorescence microscopy. The distribution of the UPIb/UPIII-negative umbrella cells was not correlated with localized urothelial proliferation (Ki-67 staining) or with the distribution pattern of CK20. Similar heterogeneities were observed in bovine but not in mouse ureter. We provide the first evidence that urothelial umbrella cells are heterogeneous as some normal-looking umbrella cells can possess only one, instead of two, uroplakin pairs. This heterogeneity seems more prominent in the urothelium of human ureter than that of bladder. This finding may indicate that ureter urothelium is intrinsically different from bladder urothelium. Alternatively, a single lineage of urothelium may exhibit different phenotypes resulting from extrinsic modulations due to distinct mesenchymal influence and different degrees of pressure and stretch in bladder versus ureter. Additional studies are needed to distinguish these two possibilities and to elucidate the physiological and pathological significance of the observed urothelial and uroplakin heterogeneity.