BACKGROUND Prognostic models for 3-year mortality after kidney transplantation based on pre-transplant donor and recipient variables may avoid futility and thus improve donor organ allocation. MATERIAL AND METHODS There were 1546 consecutive deceased-donor kidney transplants in adults (January 1, 2000 to December 31, 2012) used to identify pre-transplant donor and recipient variables with significant independent influence on long-term survival (Cox regression modelling). Detected factors were used to develop a prognostic model for 3-year mortality in 1289 patients with follow-up of >3 years (multivariable logistic regression). The sensitivity and specificity of this model's prognostic ability was assessed with the area under the receiver operating characteristic curve (AUROC). RESULTS Highly immunized recipients [hazard ratio (HR: 2.579, 95% CI: 1.272-4.631], high urgency recipients (HR: 3.062, 95% CI: 1.294-6.082), recipients with diabetic nephropathy (HR: 3.471, 95% CI: 2.476-4.751), as well as 0, 1, or 2 HLA DR mismatches (HR: 1.349, 95% CI: 1.160-1.569) were independent and significant risk factors for patient survival. Younger recipient age ≤42.1 years (HR: 0.137, 95% CI: 0.090-0.203), recipient age 42.2-52.8 years (HR: 0.374, 95% CI: 0.278-0.498), recipient age 52.9-62.8 years (HR: 0.553, 95% CI: 0.421-0.723), short cold ischemic times ≤11.8 hours (HR: 0.602, 95% CI: 0.438-0.814) and cold ischemic times 11.9-15.3 hours (HR: 0.736, 95% CI: 0.557-0.962) reduced this risk independently and significantly. The AUROC of the derived model for 3-year post-transplant mortality with these variables was 0.748 (95% CI: 0.689-0.788). CONCLUSIONS Older, highly immunized or high urgency transplant candidates with anticipated longer cold ischemic times, who were transplanted with the indication of diabetic nephropathy should receive donor organs with no HLA DR mismatches to improve their mortality risk.