Background: New onset diabetes after transplantation (NODAT) is a frequent and serious complication of renal transplantation resulting in worse graft and patient outcomes. The pathophysiology of NODAT is incompletely understood, and no prospective biomarkers have been established to predict NODAT risk in renal transplant recipients (RTR). The present work aimed to determine whether remnant lipoprotein (RLP) cholesterol could serve as such a biomarker that would also provide a novel target for therapeutic intervention. Methods: This longitudinal cohort study included 480 RTR free of diabetes at baseline. 53 patients (11%) were diagnosed with NODAT during a median [interquartile range, IQR] follow-up of 5.2 [4.1–5.8] years. RLP cholesterol was calculated by subtracting HDL and LDL cholesterol from total cholesterol values (all directly measured). Results: Baseline remnant cholesterol values were significantly higher in RTR who subsequently developed NODAT (0.9 [0.5–1.2] mmol/L vs. 0.6 [0.4–0.9] mmol/L, p = 0.001). Kaplan-Meier analysis showed that higher RLP cholesterol values were associated with an increased risk of incident NODAT (log rank test, p < 0.001). Cox regression demonstrated a significant longitudinal association between baseline RLP cholesterol levels and NODAT (HR, 2.27 [1.64–3.14] per 1 SD increase, p < 0.001) that remained after adjusting for plasma glucose and HbA1c (p = 0.002), HDL and LDL cholesterol (p = 0.008) and use of immunosuppressive medication (p < 0.001), among others. Adding baseline remnant cholesterol to the Framingham Diabetes Risk Score significantly improved NODAT prediction (change in C-statistic, p = 0.01). Conclusions: This study demonstrates that baseline RLP cholesterol levels strongly associate with incident NODAT independent of several other recognized risk factors.