AbstractThe purpose of this paper is to present a multiple-attribute error analysis in order to make more precise discrimination among competing alternatives under uncertain environments. Measurements of (especially non-monetary) attribute ratings and assessments of attribute weights are often highly imprecise and subjective, yet most analytical methods lack provisions for handling imprecise data. This paper describes how to measure the propagation of errors in the value of alternatives due to individual attribute rating and weight errors. We also measure the errors in attribute weight due to imprecise preference information from the decision-maker.