The National Cancer Institute's multicenter atypical squamous cells of undetermined significance (ASCUS) and low-grade squamous intraepithelial lesion (LSIL) triage study (ALTS) was begun to identify the most efficacious treatment protocols for women with ASCUS or LSIL. As a part of this effort, the present study investigated the sensitivity and specificity of cervicography in identifying premalignant lesions of the cervix. Study subjects were women who were referred to one of four participating clinical centers for evaluation of a diagnosis of ASCUS or LSIL on Papanicolaou (Pap) smear. Patients were randomly assigned to one of three treatment groups: immediate colposcopy, cytology only, or human papillomavirus (HPV) testing plus cytology. Cervicography was routinely performed at patient visits as a part of a safety net designed to detect any otherwise unidentified cervical cancer. For the purposes of this study, the cervigram results from the enrollment examination were compared with the cytological and histological results of the women in the immediate colposcopy and HPV testing groups. Comparison was made with 2252 ASCUS and 882 LSIL referral smears. Cervicography results were more frequently abnormal in the women with LSIL. Among women with ASCUS, cervical histology showed cervical intraepithelial neoplasia (CIN) 2 or greater in 11.6% (260 of 2252). Among those with LSIL, CIN 2 or greater was found in 20.9% (184 of 882). There were 444 final diagnoses of CIN 2 and 222 final diagnoses of CIN 3, which were compared with cervigram results at two positive test thresholds, atypical and LSIL. At the threshold of atypical, the sensitivity of cervicography would be 79.3% for detection of CIN 3. To achieve this sensitivity, it would be necessary to have a referral-to-colposcopy rate of 41.8%. At the threshold of LSIL for detection of CIN 3, the sensitivity of cervicography would be 65.8%, with 26.5% of the women referred to colposcopy. To correct for the decreased efficacy of cervicography when the entire transformation zone is not visible, the data were reanalyzed with the patients divided above and below 35 years of age. For the 2527 women younger than age 35, cervicography had a sensitivity of 80.8% for the detection of CIN 3 at the threshold of atypical. This level of sensitivity would require 47.3% of patients to be referred to colposcopy. In the 607 women 35 years old or older, cervicography had a 57.1% sensitivity for detection of CIN 3. At this level, only 19.1% of women would be referred for colposcopy.