The purpose of this study was to determine the likelihood thatintraepithelial endocervical glandular atypias that are less severe than adenocarcinoma in situ (AIS) are precursors to AIS and, if so, whether they can be recognized by morphological or other means. We first assessed the frequency of atypias found in association with either AIS or invasive adenocarcinoma (ACA) and then tested these cases and additional randomly encountered cases for the presence of human papillomavirus (HPV) and for their proliferative (G-67) index. Lesions not fulfilling the classic criteria for AIS were subdivided into high-grade (HGGA) and low-grade glandular atypias (LGGA). Atypias and controls were microdissected and tested for HPV by the polymerase chain reaction. Serial sections were labeled for Ki-67 by immunohistochemistry with the MIB-1 antibody. Eight cases (6.8%) containing glandular atypia were found in a search of 117 consecutive cone biopsy or hysterectomy specimens that also had either AIS, ACA, or both. An additional 17 cases were either randomly encountered or were received in consultation. In 3 cases, both HGGA and LGGA were present, yielding a total of 28 lesions for study. Of the 9 HGGA cases that were associated with either AIS, ACA, or CIN II/III, 6 were positive for HPV; M1B-1 reactivity in all 6 was present in greater than 25% of the nudei. Of the 3 HPV negative HGGA cases in this group, the 2 that were tested showed low MM-1 reactivity. All 3 cases of HGGA that were not associated with a diagnostic lesion were HPV negative and had low MM-1 reactivity. Of the 6 LGGAs associated with either AIS, ACA, or CIN II/III, 1 was positive for HPV, MIB-1 was nonreactive in this case and was low in all of the HPV negative cases in this group that were tested. Of 10 LGGAs not associated with a diagnostic lesion, or with a low-grade squamous lesion as the only other abnormality, 2 were positive for HPV. Of these 2, one had an MIB-1 reactivity of greater than 25% and also had intestinal differentiation. MIB-1 reactivity was elevated in 2 of the 8 HPV negative LGGAs from this group. All 10 ciliated atypias (3 HGGA, 7 LGGA) were HPV negative and had low MIB-1 reactivity. One HPVpositive AIS control case was focally ciliated. Six of 7 foci with apoptotic bodies (5 HGGA, 2 LGGA) were HIM positive. The infrequent occurrence of glandular atypias with AIS and ACA and the low rate of HPV DNA positivity when they are found in isolation are evidence that most AIS lesions do not evolve through morphologically identifiable antecedents and that most isolated atypias are not AIS precursors. HGGAs associated with AIS or CIN II/III may be either precursors to or subtle variants of AIS. However, LGGAs similarly encountered are unlikely to be either. Elevated MIB-1 reactivity may be helpful diagnostically in selected cases, but it is not reliable as an independent criterion. The presence of cilia in isolated glandular atypias favors a onneoplastic process, whereas intestinal differentiation and apoptotic bodies each suggest neoplasia.