The reliability and applicability of colposcopically directed cervical punch biopsy was assessed in a sample of 170 paired punch and large loop excision of cervical transformation zone (LLETZ) specimens obtained from previously untreated women who had been selected for treatment on the basis of cytology and/or colposcopic findings and in whom the entire cervical transformation zone was visible. A single punch biopsy was taken immediately before the LLETZ, and all the specimens were reviewed by a single pathologist. Nine (5.3%) punch biopsies were inadequate. In terms of whether or not there was cervical intraepithelial neoplasia (CIN), the chance-corrected kappa analysis rated overall agreement as poor (kappa = 0.21, 95% confidence limits 0.02-0.39), whereas in terms of histologic grade, it was fair to moderate (kappa = 0.32, 95% confidence limits 0.23-0.42). Punch biopsy tended to underestimate the disease. The sensitivity and specificity of colposcopically directed punch biopsy for the detection of high-grade CIN was 74% and 91%, respectively, with positive- and negative predictive values of 97% and 48%, respectively. Two microinvasive and two intraepithelial glandular lesions were missed on punch biopsy. Punch biopsy should be avoided when high-grade disease is suspected.