To establish the diagnostic utility of the anti-cyclic-citrullinated peptide antibody (aCCP) test in Black South Africans with early rheumatoid arthritis (RA). A cross-sectional study comparing the rheumatoid factor (RF) and aCCP status in RA patients and a control group consisting of healthy subjects, and patients with systemic lupus erythematosus (SLE) and scleroderma. The sensitivity, specificity, positive (PPV) and negative predictive values of the aCCP test alone were 82.5%, 84.9%, 87.6% and 79% versus 81.7%, 90.7%, 92.5% and 78% for RF alone. The best specificity (95.3) and PPV (95.8%) was observed when both aCCP and RF tests were positive. Patients with erosive disease had a significantly higher mean RF titre compared with those with non-erosive disease (p = 0.007). There was a trend towards an association of smoking (OR = 4.1, 95% CI = 0.9-18.6) and functional disability (p = 0.07) with RF-positive status. No similar clinical associations were observed with aCCP. Almost a third of SLE patients were aCCP positive. Despite the best specificity and PPV observed when both the aCCP and RF tests were positive, our findings suggest that testing for aCCP is only cost-effective in the RF-negative patient in whom there is a strong clinical suspicion of RA.