BACKGROUND: Circulating C-reactive protein (CRP) is associated with the metabolic syndrome and might be causally linked to it. Our aim was to generate estimates of the association between plasma CRP and metabolic syndrome phenotypes that were free from confounding and reverse causation, to assess the causal role of this protein. METHODS: We examined associations between serum CRP concentration and metabolic syndrome phenotypes in the British Women's Heart and Health Study. We then compared these estimates with those derived from a mendelian randomised framework with common CRP gene haplotypes to generate unconfounded and unbiased estimates of any causal associations. FINDINGS: In a sample of British women, body-mass index (BMI), systolic blood pressure, waist-to-hip ratio, serum concentrations of HDL cholesterol and triglycerides, and insulin resistance were all associated with plasma CRP concentration. CRP haplotypes were associated with plasma CRP concentration (p<0.0001). With instrumental variable analyses, there was no association between plasma CRP concentration and any of the metabolic syndrome phenotypes analysed. There was strong evidence that linear regression and mendelian randomisation based estimation gave conflicting results for the CRP-BMI association (p=0.0002), and some evidence of conflicting results for the association of CRP with the score for insulin resistance (p=0.0139), triglycerides (p=0.0313), and HDL cholesterol (p=0.0688). INTERPRETATION: Disparity between estimates of the association between plasma CRP and phenotypes comprising the metabolic syndrome derived from conventional analyses and those from a mendelian randomisation approach suggests that there is no causal association between CRP and the metabolic syndrome phenotypes.