Among persisting questions on bone calcification, a major one is the link between protein expression and mineral deposition. A cell culture system is here proposed opening new integrative studies on biomineralization, improving our knowledge on the role played by non-collagenous proteins in bone. This experimental in vitro model consisted in human primary osteoblasts cultured for 60 days at the surface of a 3D collagen scaffold mimicking an osteoid matrix. Various techniques were used to analyze the results at the cellular and molecular level (adhesion and viability tests, histology and electron microscopy, RT-and qPCR) and to characterize the mineral phase (histological staining, EDX, ATG, SAED and RMN). On long term cultures human bone cells seeded on the osteoid-like matrix displayed a clear osteoblast phenotype as revealed by the osteoblast-like morphology, expression of specific protein such as alkaline phosphatase and expression of eight genes classically considered as osteoblast markers, including BGLAP, COL1A1, and BMP2. Von Kossa and alizarine red allowed us to identify divalent calcium ions at the surface of the matrix, EDX revealed the correct Ca/P ratio, and SAED showed the apatite crystal diffraction pattern. In addition RMN led to the conclusion that contaminant phases were absent and that the hydration state of the mineral was similar to fresh bone. A temporal correlation was established between quantified gene expression of DMP1 and IBSP, and the presence of hydroxyapatite, confirming the contribution of these proteins to the mineralization process. In parallel a difference was observed in the expression pattern of SPP1 and BGLAP, which questioned their attributed role in the literature. The present model opens new experimental possibilities to study spatio-temporal relations between bone cells, dense collagen scaffolds, NCPs and hydroxyapatite mineral deposition. It also emphasizes the importance of high collagen density environment in bone cell physiology.