In cervical cancer screening programs, women with abnormal cytology results are referred to colposcopy for histological diagnosis. This study was designed to evaluate the sensitivity of colposcopic procedures for detecting cervical cancer and its precursor, cervical intraepithelial neoplasia (CIN). Women referred to colposcopy for abnormal cytology were enrolled from four hospitals. Gynecologists were required to take a colposcopy-guided biopsy from the worst of the abnormal-looking areas as a first biopsy. They were also asked to take ≥ 3 cervical specimens including by endocervical curettage (ECC). Random biopsies were performed at the gynecologist's discretion. We analyzed 827 biopsy results from 255 women who were diagnosed by central pathologists as having histology of CIN or cancer. In this study, 78.1% of diagnoses of CIN grade 2 or worse (CIN2+) (the threshold that would trigger intensive management) were obtained from a first colposcopy-guided biopsy. The additional diagnostic utility of second and third colposcopy-guided biopsies was 16.4 and 1.8%, respectively. The combined sensitivity of two colposcopy-directed biopsies for CIN2+ detection was >90%, regardless of the colposcopist. Random biopsies and ECC increased the diagnostic yield of CIN2+ lesions otherwise missed by colposcopy-guided biopsies alone, but only by 1.2 and 2.4%, respectively. Random biopsies were more useful for women referred after low-grade abnormal cytology (P = 0.01). The utility of ECC was greatest among women with unsatisfactory colposcopy (P = 0.03) or aged ≥ 40 years (P = 0.02). Our data suggest that at least two colposcopy-directed biopsies should be taken for histological diagnosis. Random biopsies and ECC are recommended for special populations.