Background This study aimed to assess inter-observer variability between the original diagnostic reports and later review by a specialist in breast pathology considering lobular neoplasias (LN), columnar cell lesions (CCL), atypical ductal hyperplasia (ADH), and ductal carcinoma in situ (DCIS) of the breast. Methods A retrospective, observational, cross-sectional study was conducted. A total of 610 breast specimens that had been formally sent for consultation and/or second opinions to the Breast Pathology Laboratory of Federal University of Minas Gerais were analysed between January 2005 and December 2010. The inter-observer variability between the original report and later review was compared regarding the diagnoses of LN, CCL, ADH, and DCIS. Statistical analyses were conducted using the Kappa index. Results Weak correlations were observed for the diagnoses of columnar cell change (CCC; Kappa = 0.38), columnar cell hyperplasia (CCH; Kappa = 0.32), while a moderate agreement (Kappa = 0.47) was observed for the diagnoses of flat epithelial atypia (FEA). Good agreement was observed in the diagnoses of atypical lobular hyperplasia (ALH; Kappa = 0.62) and lobular carcinoma in situ (LCIS; Kappa = 0.66). However, poor agreement was observed for the diagnoses of pleomorphic LCIS (Kappa = 0.22). Moderate agreement was observed for the diagnoses of ADH (Kappa = 0.44), low-grade DCIS (Kappa = 0.47), intermediate-grade DCIS (Kappa = 0.45), and DCIS with microinvasion (Kappa = 0.56). Good agreement was observed between the diagnoses of high-grade DCIS (Kappa = 0.68). Conclusions According to our data, the best diagnostic agreements were observed for high-grade DCIS, ALH, and LCIS. CCL without atypia and pleomorphic LCIS had the worst agreement indices. Virtual Slides The virtual slide(s) for this article can be found here: http://www.diagnosticpathology.diagnomx.eu/vs/1640072350119725.