A patient-specific cancer model is promising for identifying the most appropriate drug combinations fit to an individual person. Especially for a point-of-care-testing, an ex vivo platform for modeling patient's cancer is advantageous in efficiently testing the multiple sets of combinatorial drugs owing to the higher productivity compared to animal models. However, it has been hindered to reflect the original characteristics of parental cancer on an ex vivo experimental setting because of the inherently different environmental conditions from the in vivo ecology. Even with the current best technologies (e.g., cancers on chips and cancer organoids), it has been difficult to recapitulate the extreme heterogeneity and complexity of cancer. To overcome the limitation, we have developed a highly biomimetic brain cancer-on-a-chip by integrating 3D bioprinting technology, microfluidics, and decellularization technique. Our chip simultaneously mimicked the heterogeneous anatomical structure, the oxygen gradient, and the biochemical characteristics of brain cancer. Finally, the chips printed with the patient-derived cancer cells reproduced the differential treatment resistances and drug susceptibilities of glioblastoma patients. Therefore, we expect that this approach will allow prioritizing the drug combination candidates tailored for each brain cancer patient and also providing a treatment plan guidance to clinicians.