Objective To investigate inter-observer reproducibility in the pathologic diagnosis of breast intraductal proliferative lesions (BDPL). Methods Forty three BDPL patients were diagnosed by criterion of Page. Every specimen from each case was sorted randomly. All slides were classified as mild usual hyperplasia, moderate-severe usual hyperplasia, mild atypical hyperplasia, moderate-severe atypical hyperplasia, ductal carcinoma in situ, or ductal carcinoma in situ with invasion. Inter-observer agreement of the two groups was statistically analyzed using Kappa test. Then we compared all the diagnoses of individual pathologist with the consensus opinion confirmed by two breast pathologists to analyze the diagnostic accuracy and undue diagnosis. Results Inter-observer reproducibility of the trial group was higher than that of the control group (The total K value of 6, 3, and 2 diagnoses in the two groups were 0.289 3, 0.337 1, 0.492 8, 0.100 3, 0.150 3 and 0.340 3, respectively). When the categories were simplified, inter-observer reproducibility increased. There were still undue diagnoses of different degrees among pathologists of the trial group. Conclusion Using the same criteria is an important method to increase the diagnostic reproducibility and accuracy. More practice is needed to familiarize with these criteria.