To evaluate the accuracy of bone mineral composition determination by electron microprobe analysis (EDX) the measurements have been compared to instrumental neutron activation analysis (INAA) and chemical analysis (ICPES). Bone specimens from five femoral heads were used. The trabecular content of calcium (Ca) and phosphorus (P) was analyzed by the three different methods. The EDX method allows for a microstructural analysis of intact, methylmetacrylate-embedded, undecalcified bone and the measuring points can thus be distinctly identified centrally in each trabecula. The analysis yielded 25.8 +/- 0.7 wt % Ca and 10.5 +/- 0.1 wt % P, compared with 22.2 +/- 0.5 and 23.0 +/- 1.0 wt % Ca, and 9.83 +/- 0.21 and 10.02 +/- 0.44 wt % P for INAA and ICPES, respectively. The EDX analysis was calibrated by consecutive measurements of a hard, pressed tablet of hydroxyapatit of known content. The mean Ca content deviated with -0.38 wt % from the given content and P with -0.89 wt %. We could not verify any particular interference from the embedding procedure, however, it is possible that the relatively lower P content still may reflect this. The magnesium (Mg) concentration was 0.31 +/- 0.02 wt % by EDX and 0.26 +/- 0.02 wt % by INAA. The EDX analytical method provides a useful tool for simultaneous elemental quantification in bone. It has the advantage of permitting the use of regular bone biopsy material and thus allowing for a unique microstructural evaluation of the degree of mineralization.(ABSTRACT TRUNCATED AT 250 WORDS)