We have retrospectively studied the diagnostic and predictive value of immunohistochemical characterization of adhesion molecules (ICAM-1, CD54, VCAM-1) and HLA-DR antigen in a homogeneous clinical group of 36 patients. Between 1 January 1991 and 31 January 1993, 130 patients received a kidney transplant in our unit. Biopsies of renal allografts were only performed in asymptomatic patients who had graft dysfunction, revealed by an isolated serum creatinine increase. Available frozen samples were included in this study (n = 44). The 35 cases of acute rejection diagnosed by biopsy corresponded to mild acute rejection according to the Banff classification criteria. First, we compared the expression of HLA-DR, ICAM-1 and VCAM-1 to morphological data to determine if the immunohistochemical data improved the histopathological diagnosis when the interstitial infiltrate was mild with slight tubulitis. We also studied the phenotype of infiltrating cells with monoclonal antibodies directed against T helper cells, T cytotoxic-suppressor cells, activated T cells and macrophages. Expression on tubular epithelium and density of each type of cell was graded semiquantitatively. Expression of HLA-DR, ICAM-1 and VCAM-1 was observed on tubular epithelium and endothelium in both acute rejection and other causes of graft dysfunction, limiting its diagnostic value. Activated T cells expressing CD69-AIM (activation inducer molecule) and/or HLA-DR were frequently observed in acute rejection (24/35 (69%) and 25/35 (71%) respectively) but not in other causes of renal dysfunction. We then studied the prognostic usefulness of the immunohistochemical profile in acute rejection. Of 27 patients, 12 had a progressively decreased renal function or returned to dialysis within one year after transplantation while the other 15 had a stable graft function after at least 18 months of follow-up. In the group of bad prognosis (n = 12), corticosteroid-resistant rejection episodes were significantly more frequent (p < 0.01). In this group, nine patients had an overexpression of HLA-DR on tubular epithelium versus one patient in the group of stable graft function (chi 2c = 10.57, p < 0.002). Seven patients included in the group of bad prognosis showed tubular overexpression of both ICAM-1 and VCAM-1 versus one patient in the other group chi 2c = 6.23, p < 0.02). Moreover, patients of the first group had a significantly higher number of interstitial macrophages as compared with those who had stable graft function (chi 2c = 4.87, p < 0.01). Thus, our data show that the immunohistochemical profile studied is of little value in the diagnosis of renal allograft rejection. However, an intense tubular expression of HLA-DR and/or both ICAM-1 and VCAM-1, and a high number of interstitial macrophages are significantly related to unfavorable graft outcome.