Objective: To assess the overall effects by a meta-analysis. Data sources: Electronic searches on PubMed and Ovid Medline from their start to October 2009 were carried out. Objective: Cohort studies and secondary analysis of randomised controlled trials reporting the relative risk (RR) of recurrent cardiovascular events or death associated with C-reactive protein (CRP) obtained within 72 h from acute coronary syndromes (ACS) onset. Data extraction: Two epidemiologists independently abstracted information on study design, study and participant characteristics, level of CRP, outcomes, control for potential confounding factors and risk estimates using a standardised form. Results: A general variance-based method was used to pool the estimates of risk. Thirteen studies containing 1364 new cases identified from 9787 patients during the follow-up periods reported the risk estimates by CRP categories. Compared with the bottom CRP category (≤3 mg/l), the pooled RRs and their 95% CIs were 1.40 (1.18 to 1.67) for the middle (3.1∼10 mg/l) category and 2.18 (1.77 to 2.68) for the top (>10 mg/l) category of CRP values with a random-effects model, respectively. Another four and three studies reported the risk by unit of CRP or logarithmically transformed CRP. The pooled RRs (95% CI) were 1.49 (1.06 to 2.08) per 5 mg/l and 1.26 (0.95 to 1.69) per natural logarithm of CRP (mg/l), respectively. Conclusions: Greater early blood CRP moderately increases long-term risk of recurrent cardiovascular events or death, and may be a valuable prognostic predictor in patients after ACS.