OBJECTIVES: Colorectal cancer is the third highest incidence amongst all cancers worldwide. Biologics are increasingly used as a treatment option, and due to high associated drug cost HMOs need to minimize expenditures by choosing less costly treatment strategies. Real-world data is growing in importance in health care decision making especially in coverage and reimbursement decisions. Therefore, the objective of the study is support treatment decision making by providing evidence based on real-world data, focusing on most used biologics in metastatic CRC: bevacizumab and cetuximab. METHODS: A review of administrative claims database of Unimed São José do Rio Preto (medical cooperative responsible for 118,000 lives in São Paulo-Brazil) was conducted for patients who underwent  CRC treatment between December 2009 through January 2012. In order not to disclose confidential commercial arrangements with suppliers analysis were focused on total costs of treatment (drugs, devices/materials and room taxes). In the cases where a single patient underwent treatment with more than one biologic the analysis was performed considering the different regimens for the patient, obtaining daily costs/regimen/patient, and then converted on monthly basis. Focus was given to costs related to bevacizumab plus chemotherapy (Bev+CT) and cetuximab plus chemotherapy (Cet+CT) regimens. Also, regimens were classified into irinotecan or oxaliplatin-based. Costs were reported in Brazilian Reais (BRL1.00~USD0.48 December 2012). RESULTS: A total of 108 CRC patients were identified and regimens were 22.7% Bev+CT and 16.3% Cet+CT. Approximately 80% of both biological drugs were combined with irinotecan-based schemes. Average cost/patient/month were BRL 12,585 (SD: BRL3,588) for Bev+CT and BRL 17,178 (SD: BRL3,797) for Cet+CT. CONCLUSIONS: Results indicate potential resource savings favoring bevacizumab. If all patients treated with cetuximab were treated with bevacizumab instead, it could averagely result in savings of BRL 64,301 per month (less 26.7%). Study had limitation regarding identification of treatment line and sample size precluded identification of statistical difference between treatments.