Niches in the bone marrow (BM) regulate hematopoietic stem, progenitor and precursor cell fate and behaviour through cell-cell interactions and soluble factor secretion. Niche biology and niche-hematopoietic crosstalk are very complex processes not completely elucidated yet, due essentially to human material accessibility difficulties and limits of existing in vitro culture models. Here, we develop a 3D fully human bone marrow model recapitulating all hematopoietic and non-hematopoietic niche elements to aid further investigation of normal and pathological haematopoiesis and to move away from expensive, time-consuming and imperfect animal models. At first, we have developed a protocol to in vitro production and culture of all human main cellular non-hematopoietic components of bone marrow microenvironment. This comprehensive stroma is generated from human primary cells from femoral head, without any external cell lines contribution, and it reflects accurately the characteristics of the donor. In a 2-step culture process, mesenchymal stromal cells and endothelial cells are first selected and amplified simultaneously in an amplification medium, then they are cultured in the amplification medium as control or in a specific differentiation medium patented by the authors in conventional 2D culture plates or in 3D cultures: in spheroids, by self-organization in ultra-low-adherence plates or on a biomaterial (β-TriCalcium Phosphate), in the U-CUP perfused bioreactor. The resulting cellular stroma composition is analysed by quantifying the gene expression of markers of the osteoblastic (RUNX2, OSX, BSP), adipocytic (LPL, PPARγ, ADIPOQ) and endothelial (CD31) lineages . Moreover, specific markers of hematopoietic niche as ANG1, SCF, CXCL12, LEPR, VCAM, OPN are investigated by RT-qPCR. The spatial organization and the proportion of the different cellular stroma components are analysed by immunofluorescence (Osteoblastic cells: OSX marker, adipocytic cells: Bodipy probe, endothelial cells: CD31). The resulting tissues exhibited all compositional and structural features of human physiological BM: all stroma cells are presents and they express the main niche soluble factors. So, we explore if medullary stroma differentiated using our protocol is able to support the maintenance of HSPCs, to do this we performed coculture experiments in 2D with CD34+ cells sorted from umbilical cord blood. After 14 and 21 days of coculture, hematopoietic cells are recovered, counted and analysed by flow cytometry. Results show that our differentiated stroma is capable of maintaining, expanding and functionally regulated healthy human cord blood-derived HSPCs. An immature compartment (CD34+CD38-) is still present at day 21. Importantly, our primary in vitro niche model supports HSPCs with no cytokine addition. The same results are obtained with healthy human peripheral blood-derived HSPCs cocultured in 3D with the stroma organised in spheroids. Moreover, we demonstrated that our BM model is capable not only to maintain HSPC compartment but also to drive hematopoietic differentiation. In 2D coculture we detected mature blood cells and in 3D cultures, peripheral blood-derived HSPCs cocultured with spheroids are able to engage in erythroid lineage and recapitulate the early phases of erythropoiesis. These results demonstrate the fully complexity and huge versatility of our functional in vitro 3D model of human hematopoietic niches. Our exhaustive fully human hematopoietic niche model prepared from the primary cells of single donor, enables the design of advanced, tuneable in vitro BM proxies for the study of human physiological and pathological haematopoiesis and for screening molecules for predictive purposes in personalized medicine.