The bone is a complex and dynamic tissue, in which the equilibrium between bone deposition and resorption can be perturbed by various pathological conditions, including bone metastases. Against them, no effective therapy has been developed yet, and available treatments are primarily palliative, aiming at restoring bone homeostasis. To improve the process of anti-metastatic drug discovery, new pre-clinical models are required, since available in vivo and in vitro models are limited by species specific differences in tumor mechanisms and by an oversimplification of the bone environment, respectively. Furthermore, potential side effects can be neglected with available models, resulting into unexpected toxicity of candidate drugs in clinical trials. In this scenario, advanced 3D in vitro models could become relevant assets for research and pharma industry for the discovery of new drugs against bone tumors, overcoming limitations of current models. To this end, our work aims at developing complex 3D in vitro models of bone tissue, taking into account its heterogeneous composition, to be exploited for the test of anti-metastatic drugs. To this end, we firstly developed microfluidic devices and millimeter-scaled vascularized bone models based on osteoblasts, osteoclasts, vascular cells and mesenchymal stromal cells embedded in a 3D hydrogel loaded with hydroxyapatite nanoparticles. We demonstrated that the simultaneous presence of all cell types and of the mineral component increased the bone turnover, as compared to simpler culture conditions. Then we added immune cells and breast cancer metastatic cells, showing that tumor cells were able to colonize the bone microenvironment, particularly in the perivascular niche. Furthermore, we were able to investigate the behavior of immune cells, when in co-culture with vascular and cancer cells within a bone-like environment. To test the effects of anti-tumor drugs in our system, we added rapamycin and doxorubicin, two FDA-approved anti-tumor drugs, known to have side effects on the vascular compartment, demonstrating how tumor cell resistance to the drugs was increased by the presence of a bone microenvironment. Furthermore, we were able to show the antiangiogenic effects of the drugs, by monitoring the damage to the microvascular network in the model. Finally, we tested our 3D bone model also with cells deriving from Ewing sarcoma, a pediatric bone tumor, showing that they could proliferate in our mineralized bone model. Sarcoma cell viability was affected by the knockdown of relevant genes and by the addition of oxidative stress inducers, in accordance with results shown in mouse experiments. Overall, we showed that the screening of anti-metastatic drugs in an in vitro model recapitulating the complexity of bone environment allowed to better estimate the effects of potential drugs both on their intended target and on other components of the microenvironment as compared to simpler models.