Abstract BACKGROUND: Two-dimensional magnetic resonance elastography (2D-MRE) is an advanced magnetic resonance method with high diagnostic accuracy for predicting advanced fibrosis in non-alcoholic fatty liver disease (NAFLD) patients. However, no prospective, head-to-head comparisons between 2D-MRE and clinical prediction rules (CPRs) have been performed in patients with biopsy-proven NAFLD. AIM: To compare the diagnostic utility of 2D-MRE against that of eight CPRs (AST:ALT ratio, APRI, BARD, FIB-4, NAFLD Fibrosis Score, Bonacini cirrhosis discriminant score, Lok Index and NASH CRN model) for predicting advanced fibrosis in a prospective cohort with paired liver biopsy as the gold standard. METHODS: This is a cross-sectional analysis of a prospective study of 102 patients (58.8% women) with biopsy-proven NAFLD, 2D-MRE and clinical research assessment within 90 days of biopsy. Receiver operating characteristic (ROC) analysis was performed to assess the performance of 2D-MRE and CPRs for predicting advanced fibrosis. RESULTS: The mean (±s.d.) age and BMI were 51.3 (±14.0) years and 31.7 (±5.5) kg/m(2) respectively. 48, 26, 9, 13 and 6 patients had stage 0, 1, 2, 3 and 4 fibrosis respectively. The area under ROC curve (AUROC) was 0.957 for 2D-MRE and between 0.796 and 0.861 for the CPRs. FIB-4 was the best-performing CPR at predicting advanced fibrosis with AUROC of 0.861. In head-to-head comparisons using the DeLong test, 2D-MRE had significantly better AUROC (P < 0.05) than each CPR for predicting advanced fibrosis. CONCLUSION: Compared to clinical prediction rules, 2D-MRE provides significantly higher accuracy for the diagnosis of advanced fibrosis in NAFLD patients. © 2015 John Wiley & Sons Ltd.