False-negative cervical cytology due to sampling error is a well-recognized problem. Forty-seven women with histologically proven cervical intraepithelial neoplasia (CIN) were studied. All had two cervical smears performed at a mean interval of 3 months. At the time of the second smear, cervicography, colposcopy and biopsy were performed. The area of acetowhite cervical lesions and of the total visible atypical transformation zone (ATZ) were measured from the cervical photographs. The 17 women in whom one or both smears showed no dyskaryosis were found to have a significantly smaller proportion of their ATZ affected by CIN. It is suggested that this finding can account for the sampling error which causes false-negative cervical cytology. New screening techniques, such as cervicography, may offer a method of detecting and assessing these relatively smaller cervical lesions.