Objective To compare and evaluate the sensitivity, specificity, accuracy, negative and positive predictive values, negative and positive likelihood ratios of colposcopically directed biopsy and diagnostic cone biopsy in patients with cervical intraepithelial neoplasia. Methods We searched PubMed, CBMdisc, CMCC, CNKI, and VIP to March 2004, and Cochrane Library (Issue 4, 2003). Related journals published from 1970 to 2003 and unpublished papers were hansearched. Diagnostic studies which employed colposcopically directed biopsy or diagnostic cone biopsy and compared with golden standard (pathological diagnosis of specimens obtained through therapeutic conization or hysterectomy) were included and meta-analysis was performed. Participants were clinically suspected of pre-cancerous cervical lesions. Quality of studies was assessed, and SROC curve by Diagnostic and Screening Group of the Cochrane Collaboration was used to perform meta-analysis. Parameters were sensitivity, specificity, accuracy, predictive values, and likelihood ratio. Results Twenty six studies (3 376 patients ranging from 2 to 604 patients/per study) met the inclusion criteria. The quality of studies was generally poor.Before sensitivity analysis, superiority of diagnostic cone biopsy (sensitivity and specificity: 0.83) was shown over colposcopically directed biopsy (sensitivity and specificity: 0.76) (P＜0.001); while after sensitivity analysis the results reversed (sensitivity of diagnostic cone biopsy was 0.58 and its specificity was 0.61; sensitivity and specificity of colposcopically directed biopsy increased to 0.84) (Plt;0.001). Conclusions No definite conclusioncan be drawn as to which method is superior. To make further analysis, more studies with high quality are needed.