Abstract: The present study aimed to analyze the distribution of CD133+ cells in usual, as well as atypical, breast hyperplasia, carcinoma in situ, and invasive breast carcinomas. The relationship between the distribution and clinicopathologic features of CD133+ cells was also explored. Methods: Immunohistochemical staining was conducted to detect expression of CD133+ cells in 45 cases with normal breast tissues, 41 with usual breast hyperplasia, 39 with atypical breast hyperplasia, 51 with breast carcinomas in situ, and 121 with invasive breast carcinomas. Results: CD133+ was not expressed in the cells of normal breast tissues. The positive expression rate of CD133 increased in usual, as well as atypical, breast hyperplasia, carcinomas in situ, and invasive breast carcinomas, with progression of lesions ( P < 0.05 ). The expression rates were 31.7% ( 13/41 ), 48.7% ( 19/39 ), 64.8% ( 33/51 ), and 74.4% ( 90/121 ), respectively, with significant differences among them ( P < 0.01 ). CD133 was also upregulated with an increase in histological grade grade I: 63.6% ( 21/33 ), grade II: 72.7% ( 26/36 ), grade III: 82.7% ( 43/52 ); P < 0.05 and TNM stage stage I: 57.1% ( 12/21 ), stage II: 69.4% (34/49), stage III: 68.7% ( 11/16 ), stage IV: 94.3% ( 33/35 ); P = 0.000 . CD133 expression was higher in invasive breast cancer with lymph node metastasis or distance metastasis compared with that without metastasis ( P < 0.05 ). The positive expression rate of CD133 was higher in patients with tumor recurrence compared with that in those patients without recurrence 82.8% ( 53/64 ) vs. 64.9% ( 37/57 ), P < 0.05 . There was no significant correlation among CD133 expression and factors such as age and menopausal status of the patients, histological type, and tumor size, as well as the expression of estrogen receptor, progesterone receptor, and Her-2 ( P > 0.05 ). Conclusion: CD133+ cells possibly play important roles in breast carcinogenesis. CD133+ breast cancer cells are closely related to invasive properties and may probably be a predictor of poor prognosis.