The Bioconductor package made4 facilitates multivariate analysis of microarray data, and builds on extensive experience of multivariate data analysis in ecology. Multivariate data analysis methods provide many useful tools to extract meaningful biological information from these large data sets. Sometimes, these methods are overlooked because they are thought to be complicated and subject to barely met application hypotheses. This is partly true in the framework of the Gaussian approximation model of multivariate analysis. But the geometric model (for example Le Roux and Rouanet, 2004), and the duality diagram (Holmes, 2006) lift most of these assumptions.