A survey design which uses two tests in series greatly increases the efficiency of screening. We conducted a study to ascertain the sensitivity, specificity, and predictive value of a screening questionnaire for systemic lupus erythematosus (SLE) patients. A series design in which this questionnaire is used, followed by an antinuclear antibody test, increased the predictive value of a positive screening result by 5 to 10-fold, depending on estimates of disease prevalence. An empirical study of the series design identified two new cases of SLE and indicated that the use of a questionnaire based on accepted criteria is theoretically sound and practical. This approach should prove useful for population surveys of SLE to avoid the sampling and detection biases of previous epidemiologic studies.