The aim of this paper is to define and investigate outlier-proneness for multivariate distributions. This is done by using a concept of ordering multivariate data based on isobar-surfaces, which yields an utmost analogy of the results to the univariate case.