In recent years a growing amount of data on prognostic features of breast cancer has allowed for identification of tumors with a very low risk of recurrence. Markers used to predict the risk of distant spread include classic clinicopathologic features as well as newer tumor gene signatures, which have been validated and are being used in cohorts of patients with breast cancer patients who have low-risk disease. However, the definition of "low-risk" breast cancer requires consideration of patient-related factors such as comorbidities and age in addition to tumor characteristics, as high competing risks for mortality might be more important than cancer recurrence from a patient's point of view. In addition, identification of low-risk breast cancer cohorts is only valuable if treatment decisions are based on this information. Therefore, the magnitude of any treatment benefit in low-risk disease needs to be quantified in a comprehensible way. However, treatment benefit in low-risk situations is hard to quantify, may vary over time, and needs to be compared to individual risks for side effects. Decision models considering tumor and patient characteristics as well as predictive markers for treatment efficacy and toxicity will be needed. Tools such as Adjuvant! Online ultimately need to include information from gene signatures in order to reliably recommend specific treatment options for patients with breast cancer patients who have low-risk disease.