The objective of this investigation was to examine the effect of cemented intramedullary stem bypass on bone torsional property in the presence of femoral cortical defect. We intended to test two hypotheses; first, intramedullary fixation without bypass will accentuate the stress concentration effect and second, there will be an optimal length of stem bypass beyond the defect. Cemented intramedullary stems were used to bypass 50% diaphyseal diameter unicortical defects in paired, fresh-frozen canine femora. One member of each pair served as an unaltered control and all specimens were tested to failure in torsion. Both single-tailed paired t-test and analysis of variance were used for data analysis. Bones subjected to the cortical defect and no bypass were substantially weakened, exhibiting only 44 ± 8% of control side maximum torque (p < 0.001). Positioning of the intramedullary stem with its tip at the center of the defect provided a small degree of strength improvement, achieving 60 ± 7% of control side maximum torque. One and two diaphyseal diameter bypass, however, significantly (p < 0.01) improved bone torsional strength, resulting in 80 ± 6% and 84 ± 13% of control side maximum torque, respectively. Three diameter bypass achieved only 68 ± 8% of control side maximum torque. This significant (p < 0.05) decline in strength when compared to two diameter bypass appears to indicate that the length of stem bypass beyond the cortical defect does have an optimum, probably due to the geometric characteristics of the canine femora.