About 30% of patients with diffuse large B-cell lymphoma (DLBCL) relapse or exhibit refractory disease (r/r DLBCL) after first-line immunochemotherapy. Bone marrow (BM) involvement confers a dismal prognosis at diagnosis, likely due to the interaction between neoplastic cells and a complex tumor microenvironment (TME). Therefore, we developed a 3D in-vitro model from human decellularized femoral bone fragments aiming to study the role of mesenchymal stromal cells (MSC) and the extracellular matrix (ECM) in the adaptation, growth, and drug resistance of DLBCL lymphoma cells. The 3D spatial configuration of the model was studied by histological analysis and confocal and multiphoton microscopy which allowed the 3D digital reproduction of the structure. We proved that MSC adapt and expand in the 3D scaffold generating niches in which also other cell types may grow. DLBCL cell lines adhered and grew in the 3D scaffold, both in the presence and absence of MSC, suggesting an active ECM–lymphocyte interaction. We found that the germinal center B-cell (GCB)-derived OCI-LY18 cells were more resistant to doxorubicin-induced apoptosis when growing in the decellularized 3D bone scaffold compared to 2D cultures (49.9% +/- 7.7% Annexin V⁺ cells in 2D condition compared to 30.7% + 9.2% Annexin V⁺ 3D adherent cells in the ECM model), thus suggesting a protective role of ECM. The coexistence of MSC in the 3D scaffold did not significantly affect doxorubicin-induced apoptosis of adherent OCI-LY18 cells (27.6% +/- 7.3% Annexin V⁺ 3D adherent cells in the ECM/MSC model after doxorubicin treatment). On the contrary, ECM did not protect the activated B-cell (ABC)-derived NU-DUL-1 lymphoma cell line from doxorubicin-induced apoptosis but protection was observed when MSC were growing in the bone scaffold (40.6% +/- 5.7% vs. 62.1% +/- 5.3% Annexin V⁺ 3D adherent cells vs. 2D condition). These data suggest that the interaction of lymphoma cells with the microenvironment may differ according to the DLBCL subtype and that 2D systems may fail to uncover this behavior. The 3D model we proposed may be improved with other cell types or translated to the study of other pathologies with the final goal to provide a tool for patient-specific treatment development.