Monotonic tensile tests and uniaxial fatigue tests were conducted on devitalized human cortical bone specimens. Fatigue testing was conducted with strain rantes (Δε) from 0.005 to 0.010 and mean strains of either −0.002, 0.0, or +0.002. The stress range of the first loading cycle (Δσ₀) was documented for each specimen. The results showed that the number of cycles to failure (N_f) was independent of mean strain and that fatigue is more strongly controlled by strain range than stress range. Data scatter was much more pronounced in plots of Δσ₀ versus N_f than in plots of Δε versus N_f. Statistical analysis showed that much of the data scatter in the Δσ0 versus N_f curve could be explained by a strong positive correlation between specimen modulus and fatigue life. There was a weak negative correlation between specimen porosity and fatigue life and a weak positive correlation between N_f and bone density but no correlation between N_f and bone ash fraction. In analyzing the Δε versus N_f plot, there was a weak positive correlation between N_f and specimen modulus and a weak negative correlation between N_f and bone porosity. No significant correlations were found between fatigue life and bone density or ash fraction. The data also show that the fatigue resistance of bone is much lower than indicated by previous bending fatigue tests. The results suggest that the fatigue strength of cortical bone at 107 cycles may be closer to 7 than 40 MPa as indicated by previous bending fatigue tests.