Insufficient primary stability of screws in bone leads to screw loosening and failure. Unlike conventional continuum finite-element models, micro-CT based finite-element analysis (micro-FE) is capable of capturing the patient-specific bone micro-architecture, providing accurate estimates of bone stiffness. However, such in silico models for screws in bone highly overestimate the apparent stiffness. We hypothesized that a more accurate prediction of primary implant stability of screws in bone is possible by considering insertion-related bone damage. We assessed two different screw types and loading scenarios in 20 trabecular bone specimens extracted from 12 cadaveric human femoral heads (N = 5 for each case). In the micro-FE model, we predicted specimen-specific Young's moduli of the peri-implant bone damage region based on morphometric parameters such that the apparent stiffness of each in silico model matched the experimentally measured stiffness of the corresponding in vitro specimen as closely as possible. The standard micro-FE models assuming perfectly intact peri-implant bone overestimated the stiffness by over 330%. The consideration of insertion related damaged peri-implant bone corrected the mean absolute percentage error down to 11.4% for both loading scenarios and screw types. Cross-validation revealed a mean absolute percentage error of 14.2%. We present the validation of a novel micro-FE modeling technique to quantify the apparent stiffness of screws in trabecular bone. While the standard micro-FE model overestimated the bone-implant stiffness, the consideration of insertion-related bone damage was crucial for an accurate stiffness prediction. This approach provides an important step toward more accurate specimen-specific micro-FE models. © 2017 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 35:2415-2424, 2017.