The aim of this study was to select the best approach for screening coeliac disease patients among populations with different grades of disease prevalence. The diagnostic performance was assessed of class A and G antigliadin antibodies and class A antiendomysium antibodies in 93 consecutive outpatients with suspected malabsorption, 44 of whom (47%) had coeliac disease according to duodenal histological tests. Class G antigliadin antibodies provided the worst diagnostic values, whereas a high diagnostic validity was found for the other two tests. The positive predictive value corrected for the disease prevalence expected in coeliac disease relatives (5%) and the general population (0.2%) fell to 30% and < 2% respectively for class A antigliadin antibodies, whereas it remained 100% for antiendomysium antibodies in both situations, providing an optimal value for their use as a screening test and as a valid alternative to duodenal biopsy when this is not feasible. The high cost of anti-endomysium antibodies and the invasive nature of duodenal biopsy prevent them being used widely as screening procedures. A cost effective two step approach was simulated measuring class A antigliadin antibodies in all subjects of the target population (first step), and performing a confirmation test (antiendomysium antibodies or duodenal biopsy) only in subjects positive for antigliadin antibodies. The results show that such a procedure should be recommended only for subjects with an expected low disease prevalence--that is, 5% for coeliac disease relatives and 0.2% for the general population--as the positive predictive value was always 100% with an acceptable false negative rate (6% and 11% respectively), irrespective of which of the two confirmation tests was used. This approach avoids the use of the confirmation test in 63% and 89% of subjects respectively for the two levels of prevalence, resulting in a considerable reduction of the cost. Patients seen for suspected malabsorption with an expected high prevalence of coeliac disease should not have such a serological screening procedure. In conclusion, antigliadin antibodies are useful to screen for asymptomatic coeliac disease in non-hospital communities if antiendomysium anti-bodies are used as a confirmation test: the latter is reasonable valid alternative to duodenal biopsy.