Background There is much interest in reported associations between serum C-reactive protein and incident ischaemic heart disease. It is uncertain what this association represents. We aimed to assess the effect of confounding from a number of different sources in the Caerphilly Prospective Heart Disease Study and in particular whether the low grade inflammation indicated by C-reactive protein may be the mechanism whereby non-circulating risk factors may influence pathogenesis of ischaemic heart disease. Methods Plasma specimens collected during 1979–83 from 1395 men with sufficient sample remaining were assayed for serum C-reactive protein by ELISA. Subsequent mortality and incident ischaemic heart disease events were ascertained from death certificates, hospital records and electrocardiographic changes at 5-yearly follow-up examinations. Results There was a positive association between C-reactive protein and incident ischaemic heart disease (PPPPPPP=0·025), low childhood social class (P=0·014) and age (P=0·036). C-reactive protein was also associated positively with circulating risk factors including viscosity, leukocyte count, fibrinogen (all PP=0·0058). After adjustment for non-circulating risk factors the association with all-incident ischaemic heart disease and ischaemic heart disease death became non-significant, but the association with all-cause mortality remained (P=0·033). Further adjustment for fibrinogen however removed any hint of an increasing trend in odds for all three outcomes. Conclusion C-reactive protein levels are raised in association with a variety of established cardiovascular risk factors. Neither C-reactive protein nor the systemic inflammation it represents appears to play a direct role in the development of ischaemic heart disease.