Mathematical simulations and benchmark measurements were performed to assess the impact that normal variations in human calcium content have on in vivo K x-ray fluorescence measurements of lead in bone. Four sets of cortical bone tissue simulants were fabricated containing from 20.8% to 23.8% calcium (by weight) for measurement in a surrogate (phantom) of the human leg. The net counts detected in the coherent backscatter peak at 88.034 keV using a 109Cd source indicate a positive trend, with a variability of up to 17% over the range of assessed calcium content. Mathematical simulations confirm this trend and also demonstrate that the contribution of 87.3 keV Pb Kβ2 counts, which are unresolved in measurements, do not contribute significantly to the coherent peak at low levels of bone-lead content. Both measurements and simulations confirm that calcium is a statistically significant parameter in predicting the K-XRF response and suggest that lead levels may be over-predicted for individuals having low bone density compared to the calibration matrix. Simulations identify a 4.5% negative bias in measured lead values for each 1% increase in calcium weight percent in the bone matrix as compared to the calibration matrix. It is therefore important to accommodate this uncertainty when performing epidemiological studies of populations having a wide range of bone densities.