Long-term stability of endosseous implants depends on successful bone formation, ingrowth and adaptation to the implant. Specifically, it will define the mechanical properties of the newly formed bone-implant interface. 3D imaging during mechanical loading tests (in situ loading) can improve the understanding of the local processes leading to bone damage and failure. In this study, titanium screws were implanted into rat tibiae and were allowed to integrate for 4 weeks with or without the addition of the growth factor Bone Morphogenetic Protein and the bisphosphonate Zoledronic Acid. Samples were subjected to in situ pullout using high-resolution synchrotron x-ray tomography at the Tomcat beamline (SLS, PSI, Switzerland) at 30 keV with 25 ms exposure time, resulting in a total acquisition time of 45 s per scan, with a 3.6 μm isotropic voxel size. Using a custom-made loading device positioned inside the beamline, screws were pulled out with 0.05 mm increment, acquiring multiple scans until rupture of the sample. The in situ loading protocol was adapted to ensure short imaging time, which enabled multiple samples to be tested with short loading steps, while keeping the total testing time low and reducing dose deposition. Higher trabecular bone content was quantified in the surrounding of the screw in the treated groups, which correlated with increased mechanical strength and stiffness. Differences in screw implantation, such as contact between threads and cortex as well as minor tilt of the screw were also correlated to the mechanical parameters. In situ loading enabled the investigation of crack propagation during the pullout, highlighting the mechanical behavior of the interface. Three typical crack types were observed: (1) rupture at the interface of trabecular and cortical bone tissues, close to the screw, (2) large crack inside the cortex connected to the implant, and (3) first failure away from the screw with cracks propagating toward the screw-bone interface. Mechanical properties of in vivo integrated bone-metal screws rely on a combination of multiple parameters that are difficult to identify and separate one from the other.