The present invention relates to a method for prognosing cancer in a subject with triple negative (TN) breast cancer, whose tumors lack expression of the estrogen receptor (ER), the progesterone receptor (PR) and normal (not amplified) levels of the human epidermal growth factor receptor 2 (HER2). Methods and biomarkers are disclosed that are useful for predicting the overall survival (OS) potential of cancer in a subject with triple negative breast cancer or for predicting metastatic disease in a subject with triple negative breast cancer. For example, the method comprises detecting in a sample from a subject one or more biomarkers selected from the group consisting of ANK3, CD24, EIF1, KLF6, KRAS, KRT1, MAP2K4, SDC4, SLC2A3, STK3, TFAP2C, and WRN. An increase or decrease in one or more biomarkers as compared to a standard is prognostic of OS of TN breast cancer. Likewise, in another example, the method comprises detecting in a sample from a subject one or more biomarkers selected from the group consisting of ANG, DICER1, EIF1, and MSH6. An increase or decrease in one or more biomarkers as compared to a standard is prognostic of metastasis of TN breast cancer.