With breast cancer now being recognized as a heterogeneous disease, the concept of personalized medicine demands that the tumor of every individual be treated uniquely. This has lead to ever-expanding use of existing prognostic and predictive markers, and the search for better ones is ongoing. The classic prognostic tools such as tumor size, lymph node status, grade, hormone receptors, and HER2 status are now supplemented by gene expression-based tools such as PAM50 and MammaPrint. However, the overdependence of these tools on proliferation-related genes is a significant handicap. Although pathway-based signatures hold great promise in future breast cancer prognostication, the fact that every tumor has multiple functional pathways significantly limits the utility of this approach. Developed by the integration of estrogen receptor (ER), HER2, proliferation-related, and other genes, the Oncotype DX assay has been able to provide valuable prognostic information for ER-positive tumors. Newer molecular markers based on cancer stem cells, single-nucleotide polymorphisms (SNPs), and miRNAs are becoming available, but their importance needs to be validated. It is clear that breast cancer is a multifaceted process and that none of the tools can reliably predict a binary outcome (recurrence or no recurrence). The breast cancer community is still awaiting an ideal prognostic tool that can integrate knowledge from classic variables such as tumor size and grade with new throughput technology and principles of pharmacogenomics. Such a tool will not only define prognostic subgroups but also be able to predict therapeutic efficacy and/or resistance based on molecular profiling.