To enable clinicians to use autoantibody tests sensibly and economically, laboratories providing the test service have a responsibility to supply the test 'specifications'. This should include information about the test's validity and reliability, its normal range or range within a control population, its sensitivity and its specificity for different conditions against both control subjects and those with disorders which need to be distinguished by the test. This information will permit the development of predictive values, or better, of likelihood ratios which can then be used to generate clinically useful post-test probabilities. To date, autoantibody tests have been used to advantage by those skilled in their interpretation but often poorly by those less familiar with the area. It is appropriate now to demystify these tests by underpinning their clinical use with good operational data.