The purpose of this study was to explore the relationship between SF3B1 mutations and the prognoses of patients with breast cancer. Clinical and SF3B1 mutation data from The Cancer Genome Atlas were analyzed. SF3B1 mutations were evaluated as prognostic factors in all breast cancer patients and specific subgroups through Cox regression and Kaplan-Meier analyses. We also investigated the relationship between traditional parameters and SF3B1 mutations. Receiver operating characteristics curves were used to analyze common risk factors for their sensitivity and specificity in predicting SF3B1 mutations. SF3B1 mutations were a poor prognostic factor in luminal B and progesterone receptor (PR)-negative breast cancer (P < 0.01). Age at diagnosis and estrogen receptor (ER) status were associated with SF3B1 mutations in all breast cancers (P < 0.01) and in luminal B and PR-negative subgroups (P < 0.01). The age at diagnosis and ER status combined had a higher sensitivity and specificity for predicting SF3B1 mutations than each factor alone. SF3B1 mutations are a poor prognostic factor in luminal B and PR-negative breast cancer patients. These mutations are significantly associated with age at diagnosis and ER status. SF3B1 mutations may therefore be a novel therapeutic target for breast cancer patients.