On the basis of simultaneously sampled postmortem blood specimens from the left and right femoral veins the pre-analytical variation of methadone measurements was evaluated and compared to the analytical variation. The material consisted of a series of 27 duplicate samples from routine autopsy cases comprising mainly drug addicts. A chiral LC-MS/MS method was used for measurement of the R- and S-enantiomers of methadone and its main metabolite 2-ethyl-1,5-dimethyl-3,3-diphenylpyrrolinium (EDDP). The analytical CV% was determined to be in the range 3-4% for methadone enantiomers and 4-6% for EDDP enantiomers. The total measurement uncertainty (CV(T)) was estimated from the pre-analytical variation (CV(PA)), analytical variation proper (CV(A)), and variation related to calibration (traceability) (CV(Cal)) according to the relationship CV(T) = [CV(2)(PA) + CV(2)(A) + CV(2)(cal)](0.5). Uncertainty related to calibration concerned a component related to the purity of drug reference compound and a contribution from the production of calibrator solutions (CV(Cal)<1%). Pre-analytical sampling variation was estimated from the duplicate measurements of blood samples after subtraction of the analytical component. The pre-analytical variation amounted to a CV% of 19-21% for R- and S-methadone and 30-38% for R- and S-EDDP, i.e. considerably larger than the other components. Due to the squared addition principle, the resulting total uncertainty (CV(T)) became largely identical to the CV(PA), i.e. 19-21% for R- and S-methadone and 31-38% for R- and S-EDDP enantiomers. Accordingly, CV(T) exceeded CV(A) by a factor 5 or more. Dominance of the pre-analytical component of variation may also be likely for other compounds measured in postmortem blood samples. Thus, the width of the 95%-uncertainty interval (+/-2CV(T)) for a postmortem measurement is largely determined by the pre-analytical component of variation. This should be kept in mind when judging on the uncertainty of postmortem measurement results.