In November 1991 separate Medicare item numbers were introduced to distinguish screening from non-screening cervical smears for the purposes of monitoring the screening programme. Referring doctors are now expected to indicate the appropriate item number on the request form accompanying the Pap smear. To test compliance with this requirement, we examined 1000 consecutive request forms for cervical smears received in August 1993. We found that 22.7% had no item number indicated and that for a further 10.4% the item number given appeared to be incorrect. As the account issued by the laboratory must show an item number the ultimate responsibility for the choice of the item number rests with the laboratory. Using the guidelines supplied by Medicare, we formulated detailed criteria to classify smears as screening or non-screening, based on the patient's past smear history and/or clinical information provided by the referring doctor. Applying these criteria to cervical smears received in this laboratory in 1993 resulted in 70% being classified as screening and 30% as non-screening smears. Analysis of the cytological predictions for these smears showed substantially higher rates for all grades of abnormalities in the non-screening smears, thus lending support to the validity of our classification system. We conclude that classifying smears into screening and non-screening categories provides valuable information for statistical and quality assurance purposes but can only be useful for monitoring purposes if the criteria for assigning smears are applied consistently by all doctors and laboratories.