Application of three prognostic models in T1N0M0 breast cancer Objective: Apply three prognostic models "online" (Nothingham index (NPI), Adjuvantonline! (AO) and PREDICT used in routine oncology practice in order to stratify patients and define the use of adjuvant therapies in patients with stage I breast cancer (BC) to evaluate its correlation and overall survival (OS) in our population. Methods: We obtained patients' medical records data with invasive BC T1N0M0, treated at the Cancer Center of the Pontificia Universidad Católica de Chile, Santiago, Chile, from January 1997 to December 2003. Results: We analyzed data from 125 patients. Median age was 55 years (35 80). Most tumors were infiltrating ductal carcinoma (72.8%), estrogen receptor positive (88.8%), 80% received endocrine therapy (ET). The estimated ET and chemotherapy benefit was not significantly different according to the AO and PREDICT models (1.3% and 1% for CT, p = 0.13, 0.9% and 1% for ET p = 0.8, respectively). The estimated median OS on NPI (96%) was higher than calculated by AO (90.9%) and PREDICT (92.5%). Interestingly disease specific mortality estimated was 3%, similar to that observed (3.2%). While the estimated median OS by all models in the group of deceased patients was lower than in surviving, this difference did not reach statistical significance (p = 0.85). Conclusion: The prognostic models applied effectively predict OS in Chilean patients with T1N0M0 BC, but in this series, they do not sufficiently discriminate patients with poor prognosis. The addition of co-morbidities to AO does not alter the results.