Previous studies have shown that atypical squamous cells of undetermined significance (ASCUS) are a predictor of higher grade lesions when patients are followed up by biopsy. The purpose of this study was to examine follow-up on all patients with ASCUS to determine if the prediction of higher grade lesions seen in association with ASCUS is a function of bias in selection of patients for biopsy. The diagnosis of ASCUS, based on Bethesda System criteria, was made on 235 cases between June 1993 and December 1994 (2% of the total cases at Yale-New Haven clinics). Of these cases, 36 were biopsied and 94 were followed by cervical/vaginal smears (CVS). As has been seen in other retrospective studies, we found that 55% (20 of 36) biopsied after ASCUS on CVS showed condyloma or dysplasia. In cases with CVS follow-up instead of biopsy, only 26% (25 of 94) had significant lesions. We find that if ASCUS is evaluated by biopsy, the prediction for higher grade lesions is roughly twice that predicted by follow-up CVS. We purpose that nonstandardized management by gynecologists results, in many instances, in selection of patients with significant past history for biopsy. Consequently, the current literature finding of ASCUS as predictor of higher grade lesions should be reanalyzed, taking into account other parameters (past history, age, etc.) in the process of defining the clinical and biological implications of the ASCUS diagnosis.