To compare the accuracy of reporting split-sample ThinPrep (SSTP) and the Pap smear (PS) in cases of histologically confirmed adenocarcinoma in situ (AIS) and to ascertain the reasons for any discrepancies. Comparison of prospective PS and ThinPrep (TP) predictions with unblinded review of TP slides. In 30 paired cytologic samples, AIS was originally accurately predicted by PS in 20 (67%) and SSTP in 14 (47%). In the 16 cases in which SSTP failed to predict AIS, on review 4 were regarded as cytologic undercalls, and in the remaining 12 no diagnostic material was found on the first TP slide. In eight of these, further TP slides were made; in three, diagnostic material was present on subsequent TP slides, and in five it was not. Differences in cytologic appearances of AIS were noted between TP and PS. Even though in routine practice SSTP did not perform as well as PS in predicting AIS, it has the potential to do so. These results cannot anticipate the performance of direct-to-vial TP with respect to prediction of AIS.