Résumé : We retrospectively reviewed our experience with 45 kidney transplant recipients (KTR) that were switched from CNI to SRL, mainly for chronic allograft dysfunction (CAD) (41/45). The mean serum creatinine at switch was 2.5 +/- 0.8 mg/dl. At 1 year, patient survival was 93%. Death-censored graft survival was 67% at 1 year and 54% at 2 years. SRL was stopped because of severe side effects in 15 patients. Among these, eight patients developed 'de novo' high-grade proteinuria. Univariate analysis revealed that (1) a higher SRL level at 1 month was a predictor of SRL withdrawal due to severe side effects (P = 0.006), and (2) predictors of graft failure after SRL conversion were low SRL loading dose (P = 0.03) and a higher creatinine level at conversion (P = 0.003). In conclusion, the therapeutic index of SRL in patients suffering from CAD is narrow, with high exposure triggering serious adverse events that may mandate SRL discontinuation, while too low exposure may expose patients to under-immunosuppression and graft loss.