Transplant glomerulitis, observed in T cell-mediated and antibody-mediated rejection, is histologically characterized by intracapillary mononuclear cell infiltration. However, the prognostic value of counting various glomerular inflammatory cells during rejection has not been elucidated, which is a key step for the introduction of novel biomarkers in the clinics. We immunophenotyped glomerulitis during episodes of acute rejection in order to investigate their predictive value for transplant outcomes. To do so, we included 57 transplant biopsies of 57 renal transplant recipients with biopsy-proven acute rejection with a median follow-up of 4.2years. We determined average glomerular cell counts for T cells, B cells, Tregs, IL-17(+) cells, neutrophils and macrophages. Logistic and Cox regression was used to investigate the association of glomerular inflammatory cells with response to therapy and graft failure on a population level. We used novel time-dependent ROC curve analyses to investigate the value of glomerular inflammatory cell infiltrates for the prediction of transplant outcomes, applicable to the individual patient. We identified three cell types that were responsible for glomerulitis during rejection: macrophages, T cells and neutrophils. By quantification of glomerular macrophages, an emerging cell type associated with antibody-mediated rejection, we were able to predict the progression towards death-censored graft failure within the first 500days after the initial episode of rejection. With use of novel time-dependent ROC analyses, we propose dynamic sensitivities, specificities, positive and negative predictive values with their corresponding cut-off values for the average amount of glomerular macrophages, depending on what time after rejection death-censored graft failure needs prediction.