Purpose: To evaluate the effect of misfit at implant-level fixed partial dentures (ILFPDs) and marginal bone support on the generation of implant cracks. Materials and Methods: This in vitro study included a mechanical fatigue test and finite element analysis. A mechanical cycling loading test was performed using 16 experimental models, each consisting of two parallel implants subdivided into four groups based on the misfit and the supporting bone condition. The framework, firmly seated at implants, was dynamically loaded vertically with a force of 1,600/160 N and 15 Hz for 1 × 10⁶ cycles. Optical microscope, scanning electron microscope (SEM), and computed tomography three-dimensional (CT-3D) analyses were performed to detect impairments. Finite element models, representing the setups in the mechanical fatigue test, were used to represent the fatigue life. Results: None of the mechanical components presented distortion or fracture at the macroscopic level during the test. In a microscopy evaluation, the fatigue test revealed scratches visible in the inner part of the conical portion of the implants regardless of the groups. SEM and CT-3D analysis revealed one implant from the misfit/no bone loss group with a microfracture in the inner part of the conical interface. The simulated effective stress levels in the coronal body were higher in the misfit groups compared with the no misfit groups. The misfit groups presented effective stress levels, above 375 MPa, that penetrated the entire wall thickness. The no bone loss group presented an effective stress level above 375 MPa along its axial direction. In the no misfit group, the area presenting effective stress levels above 375 MPa in the conical connection was larger for the bone loss group compared with the no bone loss group. Conclusion: This study confirmed that implant fracture is an unlikely adverse event. A clear pattern of effective distribution greater than fatigue limit stresses could be noticed when the misfit was present. The dynamic load simulation demonstrated that the crack is more likely to occur when implants are fully supported by marginal bone compared with a bone loss scenario. Within the limitations of this study, it is speculated that marginal bone loss might follow the appearance of an undetected crack. Further research is needed to develop safe clinical protocols with regard to ILFPD. Palabras clave: CAD/CAM, finite element analysis, implant fracture, implant level, misfit