The structural consequences of bone density changes associated with lytic metastatic lesions were investigated using an experimental model of regular, lytic metastatic lesions in bone. Circular holes were drilled in the mid-diaphyseal cortex of paired adult canine femora. The region around the defect was demineralized in one bone of each pair with 0.8 N HCl. Specimens were tested to failure in four-point bending. Defect size was determined from conventional planar radiographs as the maximum apparent defect diameter divided by the periosteal diameter. Demineralization resulted in irregular defect geometries, which increased the maximum defect dimension 33% to 57% with respect to the original drill hole diameter. Demineralization resulted in additional strength reductions beyond those expected from the original drill hole alone. Despite the irregular demineralization patterns observed, strength reductions were in close agreement with those predicted from data for regular, nondemineralized holes (r2 = 0.93). The results demonstrate that irregular diaphyseal defect borders may not require more complex fracture risk predictors than can be determined from analytic and experimental studies of regular defect geometries. Our results also demonstrate that errors of over 100% can occur when measuring diaphyseal defect size from radiographs that are not optimally aligned with respect to the defect.