Factors affecting in vivo measurement precision and accuracy of 109Cd K x-ray fluorescence measurements

109Cd K x-ray fluorescence (XRF) measurement systems from two research centres were used to measure tibia lead content in a population (n = 530) of young adults. The group mean bone lead contents (+/-SEM) determined by McMaster University (n = 214) and the University of Maryland (n = 316) were 2.80 +/- 0.51 and 2.33 +/- 0.50 microg Pb/(g bone mineral) respectively. The mean difference of 0.47 +/- 0.71 microg Pb/(g bone mineral) was not significant. There was no evidence of a systematic difference between measurements from the two systems. Measurement uncertainties for the young adults were poorer overall than uncertainties for a population of occupationally exposed men. This was because obese subjects and women were included in the study. Regressions of precision against body mass index (BMI, defined as weight/height2) determined that uncertainties increased with BMI and were poorer for women than men. Measurement uncertainties (1sigma) were >8 microg Pb/(g bone mineral) for women with a BMI > 0.004 kg cm(-2). Poor-precision data affected population estimates of bone lead content; an inverse correlation was found between precision and bone lead content. A small number (0.4%) of individual measurements with poor uncertainties were inaccurate to within the precision. It is suggested that obese subjects, whose BMI > 0.004 kg cm(-2), should be excluded from 109Cd K XRF studies, as the measurement provides limited information and may be inaccurate.