To investigate and compare the accuracy and usefulness of diagnostic tests for antinuclear antibodies (ANA) a cross-sectional study of sera derived from patients admitted to the Department of Rheumatology was tested for the presence of ANA using either indirect immunofluorescence on HEp-2 cells, indirect immunoperoxidase techniques on HEp-2 cells and mouse kidney, or two commercial enzyme-linked immunosorbent assays (ELISA). The diagnostic sensitivity and predictive values of the tests were calculated and compared. The accuracy of tests was compared using receiver-operating characteristics (ROC) methodology. All ANA-positive sera were further analysed for the presence of antibodies against extractable nuclear antigens (anti-ENA) and anti-DNA. A moderate to good agreement was found between tests, with kappa ranging from 0.469 to 0.659. Highest sensitivity for systemic lupus erythematosus (SLE; 93.3%) and primary Sjögren's syndrome (SS; 70%) was found using immunofluorescence on HEp-2 cells. Immunofluorescence on HEp-2 cells performed statistically better than the other tests in predicting SLE but not SS. All tests except mouse kidney showed good and comparable performance in detecting sera with anti-ENA and anti-DNA. At the given cut-off values indirect immunofluorescence on HEp-2 cells performed best. All assays except mouse kidney showed performance characteristics sufficient for use in routine analysis of ANA.