Using the National Summary of Hospital Cancer Registry 2007, Japan, (HCRJ) from 219 local core cancer hospitals (LCCH) that had registered more than 29 breast cancers, we validated the Japanese classification of breast cancer (JCBC). In JCBC, most invasive ductal carcinomas (IDC) are subclassified as papillotubular carcinoma (coded as 850031 in the HCRJ) or scirrhous carcinoma (850033). Because of the confusing criterion that IDC with substantial ductal carcinoma in situ (DCIS) is papillotubular carcinoma, pathological T (pT)1 might be overestimated as pT2–3 by measuring the tumor size to include DCIS at LCCH where papillotubular carcinoma is diagnosed correctly. The LCCH were divided based on the difference between the proportion of papillotubular carcinoma to scirrhous carcinoma (PPS), that is, the proportion of 850031 cases to the sum of 850031 and 850033 cases at each LCCH (mean: 45.6%), and the PPS of the LCCH whose in-house histological classification was the origin of JCBC (standard PPS [StPPS]: 42.3%), into G5 (PPS within StPPS ± 5%), L5 (PPS < StPPS–5%), HL (StPPS + 15% ≥ PPS > StPPS + 5%), and HH (PPS > StPPS + 15%). On pT2–3, the proportion of N1–3 cases to N0 in G5 and HL was significantly lower than that in L5. The averages of the proportion at each LCCH of G5 and HL were also significantly lower than that of L5. Meanwhile, on pT1, the proportions and averages were not significantly different among the groups. The frequent overestimation of pT in G5 and HL explains their lower frequency of lymph nodal metastasis on pT2–3, leaving the frequency on pT1 unchanged. The JCBC has spoiled the accuracy of pTNM. (Cancer Sci 2011; 102: 1597–1601)