The medical records and histopathology of 250 patients who underwent cervical conization between January 1979 and December 1982 were reviewed. Two hundred thirty (92%) had endometrial curettage at the time of cervical conization. Abnormal findings were present in 7 (3%) of the 230 curettings. Limiting the performance of endometrial curettage at the time of conization to patients meeting specific criteria would have reduced the number of combined procedures by 75% without jeopardizing our ability to detect significant pathology. Endocervical curettage was performed on 221 (88%) of the 250 patients undergoing conization. The endocervical curettings were not a good predictor of the involvement of the cone margins with neoplasia. Only 7 (20%) of 35 patients with involved endocervical margins had a positive endocervical curettage. While the cone margins predicted residual carcinoma in the hysterectomy specimen with a sensitivity of 1.0, the sensitivity of the endocervical curettings for predicting residual carcinoma was 0.5. Although endocervical curettings can detect an invasive cancer not detected in the cone specimen, a negative endocervical curettage does not rule out invasive cancer above the excision line.