stratified based on if one versus multiple cervical sites were sampled (Table 1). There was more disease in quadrants with visible lesions, but a substantial proportion of disease was identified in quadrants without visible lesions.By obtaining a punch biopsy from cervical quadrants with and without visible lesions we estimated the amount of disease typically missed when biopsies are restricted to visible lesions. Approximately one-third of all CIN2 + were from quadrants without visible lesions, analogous to results by Pretorius et al.[4] who reported that 37% of all CIN2 + lesions in HIV-negative women were from colposcopically normal-appearing areas on the cervix. We recommend that screening studies include systematic sampling for histopathology regardless of the screening result, to minimize verification bias in cervical cancer screening trials.