Human epidermal growth factor receptor 2-positive (HER2+) breast cancers have been historically considered an aggressive entity with high rates of recurrence and poor survival. However, during the last 20 years, there has been a dramatic change in prognosis due to the incorporation of different anti-HER2 therapies into the neo/adjuvant chemotherapy backbone. Neoadjuvant dual blockade with trastuzumab and pertuzumab has become the standard of care for women with stage II and III HER2+ breast cancer. Trastuzumab emtansine (T-DM1) has been shown to improve outcomes if pathological complete response (pCR) is not achieved, and adjuvant extended therapy with neratinib has increased disease-free survival (DFS) and may have an impact in central nervous system (CNS) recurrences. However, these agents are both toxic for individual patients and costly for the overall healthcare system, and there are still patients that experience recurrence despite therapy improvements. At the same time, it has been shown that some patients with early-stage HER2+ breast cancer can be effectively treated with less intensive systemic therapy, using only taxane and trastuzumab, or that the chemotherapy backbone can be omitted completely. The current challenge is to properly identify which patients can receive a de-intensified regimen and which need new intensification strategies. Tumor size, nodal status, and pCR achievement after neoadjuvant treatment are well-known risk factors that can aid in making clinical decisions, but they do not accurately predict all patient outcomes. Various biomarkers have been proposed to better characterize the clinical and biological heterogeneity of HER2+ breast cancer. Immune infiltration, intrinsic subtype, intratumoral heterogeneity, and dynamic changes during treatment have been described as important prognostic and/or predictive features. The integration of all these factors will be key in the proper identification of the true risk, and individualized treatment strategy, for each patient.