OBJECTIVE: Only a few consistent findings on individual foods or nutrients that influence breast cancer risk have emerged thus far. Since people do not consume individual foods but certain combinations of them, the analysis of dietary patterns may offer an additional aspect for assessing associations between diet and diseases such as breast cancer. It is also important to examine whether the relationships between dietary patterns and breast cancer risk are consistent across populations. METHODS: We examined the risk of breast cancer with two dietary patterns, identified as "Vegetables" (VEG) and "Pork, Processed Meat, Potatoes" (PPP), common to all cohorts of the DIETSCAN project. During 7 to 13 years of follow-up, three of the cohorts--the Netherlands Cohort Study on diet and cancer (NLCS), the Swedish Mammography Cohort (SMC), and the Ormoni e Dieta nella Eziologia dei Tumori (Italy-ORDET)--provided data on 3271 breast cancer cases with complete information on their baseline diet measured by a validated food frequency questionnaire. RESULTS: After adjustment for potential confounders, VEG was not associated with the risk of breast cancer across all cohorts. PPP was also not associated with the risk of breast cancer in SMC and ORDET, but a high PPP score tended to be inversely associated with breast cancer in the NLCS study (RR = 0.69; 95% CI, 0.52-0.92, highest versus lowest quartile). PPP differed in one aspect between the cohorts: butter loaded positively on the pattern in all cohorts except NLCS, in which butter loaded negatively and appeared to be substituted by low-fat margarine loading positively. CONCLUSION: In general, the dietary patterns showed consistent results across the three cohorts except for the possible protective effect of PPP in the NLCS cohort, which could be explained by a difference in that pattern for NLCS. The results supported the suggestion derived from traditional epidemiology that relatively recent diet may not have an important role in the etiology of breast cancer.