The widespread uae of colposcopy in the evaluation of abnormal cervical cytology has been accompanied by a trend toward decreased utilization of Papanicolaou smears and cone biopsy. But its accuracy is dependent on the criteria determining its limitations and the observer's experience. This study was performed to evaluate the diagnostic accuracy of Papanicolaau smear and eolposcopic finding and to campare these findinga with thase af colposcopically directed biopsy, conization and hysteretomy and to determine the value of endocervical curettage. Nine hundred twenty six patients were evaluated from Jan. 1, 1986 to Dec. 31, 1993 at the cancer detectinn center of the department of obstetrics and gynecology, Yonsei University, College of Medicine. Three major classes were involved in this study ; 662 cases with within normal limit Papanicolaou amears with cervical examination grossly suspicious for cancer, 187 cases with squamous intraepithelial lesion or greater, and 72 caees with reactiue or reparative change. The patients with inveaive rnalignancies of the cervix visihle to the naked eye were not included in this study. The majority of patients(75%) were 31 to 50 years of her age. The colposcopic examination wee deemed unsatisfactory in 45 cases in whom the upper limit, of the transformation zone could not be visualized. There were 881 cases each with a satisfactory colposcopie examination. We obtained the following results. In the satisfactory group with within normal limit Pap smears which had clinical impression of gross suspicion for cancer, 54 of 662 cases(8%) were diagnosed as more than cervical intraepithelial neoplasia I and 36(5.5%) were diagnosed as more than cervical intraepithelial neoplasia II with the histology of colposcopically directed biopsy. The analysis of the diagnostic values of Pap smears for cervical cancer screening showed that the sensitivity was 80.0%, the specificity was 93.8%, the false negative rate was 80.0%, the false positive rate was 15.2%, the negative predicitive value was 92.1%, and the positive predicitive value was 83.6%. In the satisfactory group with within normal limit Pap smears which had clinical impression of gross suspicion for cancer, 54 of 662 cases(8%) were diagnosed as more than CIN I and 36(5.5%) were diagnosed as more than CIN III with the histolgy of colposcopically directed biopsy. The accuracy of colposcopic impression when compared to histology of the colposcopically directed biopsy in the satisfactory group was 89.8% (within 1histologic degree) and 82.7% (precise agreement). Histologic comparision of colposcopically directed biopsy and final surgical specimen showed accuracy of 87.2% (within 1 histologic degree) and 66.5% (precise agreement). The accuracy of endocervical curettage with colposcopically directed biopsy when compared with final surgical specimen in unsatisfactory group was 82.2% (with 1 histologic degree) and 57.7% (precise agreement). In conclusion, satisfactory colposcopic evaluation in conjuction with Pap smear offers accurate method for cervical cancer detection. But endocervical curettage with colposcopically directed biopsy appears to be of less value in unsatisfactory group. Therefore futher evaluation is need.