Background: Early recognition of rheumatoid arthritis (RA) is important as aggressive anti-rheumatic therapy could prevent (irreversible) joint destruction in patients with early RA. Autoantibodies against cyclic citrullinated peptide (anti-CCP) have been proposed as an important marker for the diagnosis of RA. The accuracy of these antibodies is unknown. Objectives: To determine the diagnostic accuracy of anti-CCP antibodies for the diagnosis of RA. Methods: The Cochrane guidelines for reviews of test accuracy were followed. Databases were searched until August 2007 for studies that evaluated the accuracy of anti-CCP antibodies for the diagnosis of RA. Studies were included if sufficient data were available to construct a 2x2 table and if the 1987 American College of Rheumatology criteria were used as reference standard. Two reviewers independently extracted data and assessed the methodological quality of the included studies using QUADAS criteria. Data were extracted as 2x2 tables presenting the index (anti-CCP) test results and disease status (RA versus non RA). The HSROC and bivariate random-effects models were used to pool paired sensitivity and specificity. Heterogeneity was investigated through relevant subgroup analyses. The effects of study design (case-control versus cohort), inclusion of unclassified arthritis (UA) (patients in whom a diagnosis of RA could not be ruled in or out) in the analysis, disease duration, index test characteristics, and QUADAS criteria were investigated. Results: Based on initial results, 73 studies were included in the review (26 diagnostic cohort and 47 diagnostic case control studies). The table presents pooled sensitivity and specificity for different groupings of studies. Specificity was consistently high; there was considerable variation in sensitivity across all subgroups investigated. The figure shows results from all studies plotted in SROC space together with an SROC curve and the summary point with confidence and prediction intervals around this point. Further investigation of heterogeneity in sensitivity is proposed in order to determine the group of patients in whom anti-CCP will be most clinically useful. Conclusions: Anti-CCP antibodies have excellent specificity and are able to accurately rule in a diagnosis of RA. Sensitivity showed considerable heterogeneity.