Background. Kidney biopsies are being used to evaluate marginal donors, but rigorous statistical validation of this practice with multivariate analysis has not been performed. Methods. To analyze histologic parameters in 78 donor biopsies for their ability to predict graft dysfunction, we used a proportional odds model that included both donor and recipient factors. Glomerulosclerosis was categorized into grades 0, 1, 2, and 3, corresponding to 0, 1-10%, 11-20%, and 21-30% global sclerosis, respectively. The degrees of interstitial fibrosis, tubular atrophy, arteriosclerosis, and arteriolar hyalinosis were graded from 0 to 3+, using definitions suggested by the Banff Schema of allograft pathology. Results. Increasing donor age was associated with higher glomerulosclerosis, tubular atrophy, and arteriosclerosis. Kidneys with any degree of interstitial fibrosis were 2.6 times [odds ratio (OR)] more likely to experience a worse outcome at 6 months (P=0.02). This association held up after correction for acute rejection (OR 2.5, P=0.03) and high panel-reactive antibody (OR 3.4, P=0.006). However, the OR was reduced to 1.9 (P=0.15) after controlling for recipient age. With each increment in the grade of glomerulosclerosis, the odds for a worse outcome at 12 months increased to 2.3 (P=0.005). The value for OR became 2.0 (P=0.03) when controlling for recipient age (P=0.01), 2.4 (P=0.005), when controlling for acute rejection, and 2.3 (P=0.006) when controlling for high panel-reactive antibody. Conclusions. Histopathological parameters present in donor biopsies can independently predict posttransplant graft function. Implications for the pool of donor organs available for transplantation are discussed.