Risk for new-onset diabetes (NOD) after renal transplantation is higher with tacrolimus (Tac) than with cyclosporine (CsA), but the extent to which the diabetogenic effect of Tac is dosage dependent or steroid dependent remains uncertain. Patients who received a transplant between 1995 and 2002 were drawn from the United Network for Organ Sharing registry and prescription records and NOD diagnoses from Medicare claims, both provided by the United States Renal Data System. Patients were divided into six groups of steroid and Tac doses at 30 d after transplantation and referenced against CsA. Relative hazards of NOD with Cox proportional hazards regression were estimated incorporating propensity scores for Tac and nonimmunosuppressive factors related to NOD. A total of 8839 patients with valid immunosuppression records and without pretransplantation evidence of diabetes were included in the study. Unadjusted, cumulative, NOD incidence 1 yr after transplantation was 14.6% with CsA and 22.2% with Tac and at 3 yr after transplantation was 23.4% with CsA and 32.9% with Tac (P < 0.0001). Neither higher CsA nor higher steroid dosages were associated with NOD in CsA-treated patients. However, NOD hazard was significantly higher with Tac than with CsA in all six steroid/Tac dosing groups, including the cohort with the lowest dosages of Tac (dosage thresholds at 30 d after transplantation <0.12 mg/kg per d [mean 0.07 mg/kg per d] and steroids (<0.75 mg/kg per d; hazard ratio 1.28; 95% confidence interval 1.10 to 1.48; P = 0.0012). Whereas the incidence of NOD is greatest with high Tac dosages, the increased risk versus CsA is sustained with lower Tac dosages. Higher steroid dosages increase the early diabetogenic effect of Tac but not of CsA.