Rheumatoid factor is a commonly requested test in the investigation of joint problems. Although the test has been found to be both sensitive and specific for rheumatoid arthritis in patients attending Rheumatology Clinics, when the test is applied to general hospital or community populations it performs poorly. Thus the setting in which rheumatoid factor tests are requested and awareness of the test's limitations are critical for appropriate use and interpretation of results. We studied 295 consecutive requests for testing at four centres. The majority were performed for the investigation of joint problems but only 6% of these were positive, suggesting unselective requesting. Half of the results were used to make diagnoses and influence management. However, significant numbers of clinicians felt that the test result had either excluded or confirmed a diagnosis of rheumatoid arthritis suggesting that results were being over-interpreted and that clinical decisions made on this basis may have been inappropriate. We conclude that in this setting rheumatoid factor testing is of limited value and generates misleading information.