Of the approximately 60 million Pap smears performed in the United Sta
tes in 1995, about 8% or 5 million will show cytology that is ''not ne
gative'' (ASCUS, AGCUS, LSIL, HSIL, etc.). Possibly 15% or about 0.7 m
illion of these cases will have positive follow-up by repeated Pap sme
ars, colposcopy or biopsy. More than 4 million will be false-positive
smears based on the reference standard of biopsy or repeated smears. I
f no treatment or medical intervention was offered to the 0.7 million
cytologically and histologically positive cases, perhaps 20,000 (3%) w
ould develop into invasive cancer. Of the original 5 million cytologic
ally ''not negative'' cases, fewer than 0.5% have the potential to dev
elop into invasive cancer. While considerable attention has been paid
to false-negatives in Pap screening, the above considerations indicate
that the cytological and histological criteria for assessing the mali
gnant potential of ''not negative'' samples might benefit from some re
finement. Until such refinement occurs, any chemoprevention studies in
cervix face a formidable signal-to-noise problem-worse than 1:30. Thi
s paper presents data from quantitative image cytometry of cervical sm
ears for assessing the malignant potential of various ''not negative''
cases. We have approached this in two ways-by analyzing dysplastic ce
ll nuclei and by analyzing the nuclei of cytologically normal cells gr
owing in the vicinity of the neoplastic lesion. In both cases, nuclear
features describing the distribution of the DNA in the cell nuclei (e
specially texture features) are the discriminating factors. Future res
earch into the objective assessment of malignant potential of ''not ne
gative'' cases is outlined. (C) 1995 Wiley-Liss, Inc.