Aims/hypothesis An accurate molecular diagnosis of diabetes subtype confers clinical benefits; however, many individuals with monogenic diabetes remain undiagnosed.Biomarkers could help to prioritise patients for genetic investigation. We recently demonstrated that highsensitivity C-reactive protein (hsCRP) levels are lower in UK patients with hepatocyte nuclear factor 1 alpha Diabetologia (2011) 54:2801-2810 DOI 10.1007/s00125-011-2261 (HNF1A)-MODY than in other diabetes subtypes. In this large multi-centre study we aimed to assess the clinical validity of hsCRP as a diagnostic biomarker, examine the genotype-phenotype relationship and compare different hsCRP assays. Methods High-sensitivity CRP levels were analysed in individuals with HNF1A-MODY (n=457), glucokinase (GCK)-MODY (n=404), hepatocyte nuclear factor 4 alpha (HNF4A)-MODY (n=54) and type 2 diabetes (n=582) from seven European centres. Three common assays for hsCRP analysis were evaluated. We excluded 121 participants (8.1%) with hsCRP values >10 mg/l. The discriminative power of hsCRP with respect to diabetes aetiology was assessed by receiver operating characteristic curvederived C-statistic. Results In all centres and irrespective of the assay method, meta-analysis confirmed significantly lower hsCRP levels in those with HNF1A-MODY than in those with other aetiologies (z score −21.8, p<5×10 −105 ). HNF1A-MODY cases with missense mutations had lower hsCRP levels than those with truncating mutations (0.03 vs 0.08 mg/l, p<5× 10 −5 ). High-sensitivity CRP values between assays were strongly correlated (r 2 ≥0.91, p≤1×10 −5 ). Across the seven centres, the C-statistic for distinguishing HNF1A-MODY from young adult-onset type 2 diabetes ranged from 0.79 to 0.97, indicating high discriminative accuracy.