New percutaneous filling techniques are beginning to be used in bone tumour pathology, because they are less aggressive than surgery. The purpose of this study was to test percutaneous injectable bone biomaterials with a reproducible model. A closed cancellous bone defect was created in the distal femoral extremities of 34 rabbits. Filling was done by a percutaneous injection made on the medial side, after the defect had been hollowed by a lateral approach. Three situations were tested: unfilled, filled with orthopaedic cement, filled with a soft collagen-hydroxyapatite material. Three time intervals (2, 4 and 8 weeks) were tested for the three situations. For the mixed hydroxyapatite-collagen material, each component was tested separately: the injectable collagen, and the hydroxyapatite powder. The quality of the defect was assessed by the variations of the defect area on sagittal sections. Bone formation in the defect was quantified for each group and time interval. A bone defect of reproducible size was obtained. Evolution of bone formation was different in each group. An unfilled defect was never completely filled by bone and the defect bone formation rate stayed between 9.9% and 15.1% without any statistical difference between the time intervals. The percutaneous injection of orthopaedic cement, which was progressive, was less complete than an opened surgical filling and explained a frequent low peripheral bone formation. A lower bone formation rate was observed in all the filled groups (orthopaedic cement, hydroxyapatite-collagen, hydroxyapatite and collagen) than in the control unfilled group. This study showed the use of bone biomaterials by an injectable percutaneous method, and a model to evaluate these materials is proposed.