OBJECTIVE: Renal dysfunction predicts an increased risk of both early and long-term mortality after cardiac surgery. Cystatin C enables glomerular filtration rate (GFR) to be estimated accurately and may be superior in this regard to creatinine-based estimates. We hypothesised, therefore, that cystatin C and derived estimates of GFR would independently predict long-term survival after cardiac surgery and would be superior in this respect to traditional estimates of GFR. The current study tests this hypothesis in a large and well-characterised cohort of patients. DESIGN: A prospective cohort study. SETTING: Regional cardiothoracic centre in Northeast Scotland. PARTICIPANTS: 1010 patients undergoing non-emergent cardiac surgery between 2004 and 2007. Serum creatinine and cystatin C levels were measured preoperatively and demographic and clinical variables were recorded. PRIMARY OUTCOME MEASURE: All-cause mortality, established from the National Records of Scotland. RESULTS: The median duration of follow-up after surgery was 9.7 years (IQR 8.9-10.6 years), during which 297 participants died. Preoperative creatinine and cystatin C levels and estimates of GFR derived from these were all strong predictors of death using Cox regression and remained independently predictive after adjustment for the logistic European System for Cardiac Operative Risk Evaluation, a well-validated clinical risk score and a range of other clinical predictors. Cystatin C-based measures were superior to creatinine-based estimates of GFR. CONCLUSIONS: Cystatin C and creatinine derived eGFR are powerful and independent predictors of long-term mortality following cardiac surgery. Estimates of GFR derived from cystatin C convey superior prognostic information to conventional creatinine-based estimates, but the observed differences are modest.