A 109Cd K x-ray fluorescence (XRF) system using a point source in a back-scatter geometry is described. The suitability of plaster-of-Paris phantoms as targets for intercalibration standards was evaluated. When the phantom concentrations were measured by inductively-coupled-plasma mass spectrometry (ICPMs), the calculated phantom concentrations underestimated true concentrations by an average of 15%. Since calculated values are used to calibrate the K XRF system, in vivo bone-lead concentrations may be similarly underestimated. The difference between calculated and measured concentration is attributable to impurities in the plaster of Paris (e.g. calcium carbonate). The ICPMS-measured concentrations were used to calibrate the K XRF system. The same phantoms were also measured as 'unknowns' by a bone-lead measurement system (Abiomed, Danvers, MA, USA). The commercial system overestimated the lowest-concentration phantoms and underestimated the phantoms with concentrations above 15 microg Pb/g plaster of Paris. The commercial system and our system were compared by measurement of the new phantoms in air and in water. The K XRF system exhibited better precision in both situations. On the basis of this work, we recommend that plaster-of-Paris phantoms used to calibrate K XRF measurement systems be analysed first by ICPMS or another valid analytical technique.