Abstract Rollins et al. (1995, Powder Technol. 84, 277–282), showed the superiority of the analysis of variance technique (ANOVA) over the mixing index approach in the detection of segregation in binary powder mixtures. In industry, a common problem is determinating the level of segregation in a multicomponent mixture. This article compares the performance of multivariate analysis of variance (MANOVA) technique to that of multivariate indices. Using a Monte-Carlo simulation study, the techniques performances are evaluated for varying levels of segregation, between species (intra) correlation, level of measurement error and level of test significance.