To address quality requirements for breast pathology in the Australian screening programme, one breast cancer Screening and Assessment Service initiated a process of central pathologic review of all lesions detected through the service. The aim of this study was to measure concordance between the initial and review pathology, and to assess the merit of routine review. Concordance was measured by observed agreement and the kappa statistic for 267 women with 273 lesions. Concordance was excellent for the four classification schemes examined, good for the identification of benign lesions and hyperplasia, and excellent for the identification of DCIS or invasive carcinoma. For the sub-categorization of hyperplasias and invasive carcinomas concordance was good, but was poor for the sub-typing of DCIS. Initial and review concordance was acceptable, suggesting that disagreement among pathologists may not present a major impediment to the provision of dependable diagnoses. Full case review is unnecessary for benign lesions or invasive carcinoma, but should be maintained for DCIS and hyperplasias. (C) 1999 Harcourt Publishers Ltd.