In conventional analytical chemistry it is customary to report figures of merit – such as precision, analysis time, limit of detection – for the chemical analysis performed. Usually, such figures of merit are reported for each analyte separately, generating a list of figures of merit. In metabolomics such a listing is not informative, since very many compounds are measured. An ANOVA-based strategy is proposed for a global measure of precision of the whole experiment, broken down in components of variation contributing to the total variation. This strategy uses well established statistical techniques and can be used easily. It was implemented to study the reproducibility of different comprehensive GC- and LC-MS methods for the analysis of tobacco aerosols. The results give insight into different sources of variation contributing to the total variation, such as biological variability, sampling variability and repeatability. For the specific example, median CV values ranged from 4.6% to 12.5% for repeatability; from 14.7% to 18.0% for sampling variability and, finally, from 24.2% to 26.8% for biological variability. Such a breakdown of sources of variability gives clues for improving the methods.