Recent studies have highlighted the role of androgen receptor (AR) as a prognostic biomarker of breast cancer. However, its predictive role in disease free survival (DFS) and overall survival (OS) still remains inconclusive. The present study aimed to retrospectively investigate the association between AR and survival outcomes in breast cancer and also identify this association by a meta-analysis of published researches. Clinical data from 109 patients with breast cancer, who underwent surgery at Ruijin Hospital, Shanghai, were retrospectively analyzed for immunohistochemical AR expression measured by tissue microarray. For meta-analysis, articles available in Pubmed on the relationship between AR and breast cancer outcomes were included. Data obtained from both were combined and analyzed. Women with AR positive tumors in the retrospective study had a significantly better DFS (HR 0.24, 95% CI 0.07-0.88) and OS (HR 0.19, 95% CI 0.04-0.85) than women with AR negative ones. Meta-analysis showed that AR expression in breast tumors was an indicator of better DFS (HR 0.52, 95% CI 0.43-0.64). In subgroup analysis, AR could predict DFS outcome in estrogen receptor (ER) positive (HR 0.45, 95% CI 0.34-0.59), ER negative (HR 0.42, 95% CI 0.26-0.67), and triple negative breast cancer (HR 0.40, 95% CI 0.23-0.69). Moreover, in ER positive breast cancer patients, the expression of AR could predict better OS (HR 0.39, 95% CI 0.19-0.82). The present analysis indicated that AR expression was associated with lower risk of recurrence in patients with all breast cancer types and better OS in cases with ER positive.