The authors investigated the association between mammographic parenchymal patterns and breast cancer risk by a case control study. Three-hundred and thirty-four cancers, either screen or interval detected in the course of a screening program, were assumed as cases and compared to 668 non-cancer controls matched by age, residence and date of screening examination. Parenchymal patterns were classified by retrospective blind review of randomly mixed films by two independent readers. A high agreement was achieved on a two-group scale (N1-P1 vs P2-DY) inter and intraobserver consistency being 0.94 and 0.95, respectively. The relative risk of breast cancer exposure (RR) with respect to N1 pattern was 1.99 for P1, 3.59 for P2 and 4.18 for DY patterns. The RR of P2-DY with respect to N1-P1 patterns was 2.62. The results did not change after stratification by age group. Parenchymal patterns do not seem to be a reliable criterion for selecting high risk women for screening as in this series only 57% of all cancers occurred in high risk (P2-DY) subjects.