Colposcopy and colposcopically directed biopsies were performed in 930 patients with an abnormal Pap smear; 898 patients had satisfactory colposcopic evaluation and 32 patients had unsatisfactory colposcopy, including patients suspected of having invasive carcinoma. The colposcopic impression was accurate when compared to the histology of the colposcopically directed biopsy in 91% of patients. There were 268 patients who had a cone or hysterectomy for the treatment of advanced cervical intraepithelial neoplasia. Evaluation of final histologic specimen in this group revealed a high correlation with colposcopic impression and colposcopically directed biopsy, with misdiagnosis of only three patients with early microinvasive carcinoma. Thirteen cases of occult invasive carcinoma were diagnosed by cone biopsy, five patients had colposcopic impression suggestive of invasive carcinoma, and eight had unsatisfactory colposcopy. No occult or frank invasive carcinoma was missed in those patients who had satisfactory colposcopy. In those patients who had unsatisfactory colposcopy, diagnostic cold-knife conization was performed. Final histologic diagnosis disclosed a more advanced lesion in over 50% of this group. Omission of endocervical curettage from the colposcopic examination of 798 patients with satisfactory colposcopy did not reduce the accuracy of the procedure. The results of this study confirm those of other investigators that satisfactory colposcopic evaluation is a highly accurate method for evaluating abnormal cytologic smears.