Objective To evaluate the diagnostic value of antikeratin antibody (AKA) for rheumatoid arthritis (RA). Methods Systematic and comprehensive literature was searched in PubMed (1966 to June 2010), The Cochrane Library (Issue 6, 2010), CBM (1978 to June 2010), CNKI (1994 to June 2010), VIP (1989 to June 2010), and CMA Digital Periodicals (1997 to June 2010). The diagnosis studies of antikeratin antibody for rheumatoid arthritis were included. The quality assessment of diagnostic accuracy studies (QUADAS) items were used to assess the quality of the included studies. The Meta-Disc (version 1.4) software was used to analyze the data. Results A total of 69 trials involving 14 890 participants were included. The results of meta-analyses showed that compared with the RA classification criteria revised by American Rheumatism Association (ARA), the summary sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, OR value, and summary receiver operating characteristic curve of antikeratin antibody were 0.41 (0.39, 0.42), 0.94 (0.94, 0.95), 9.52 (7.21, 12.57), 0.63 (0.60, 0.66), 15.24 (11.62, 19.98), and 0.613 6, respectively. Conclusion Antikeratin antibody might be one of the most effective diagnoses for rheumatoid arthritis. The clinicians should combine other autoantibodies with AKA to diagnose rheumatoid arthritis.