DIVCLUS-T is a descendant hierarchical clustering method based on a monothetic biclustering approach allowing the dendrogram to be read as a decision tree. Like in Ward ascendant method, the within-cluster inertia is minimized at each iteration. The DIVCLUS-T constructs then a indexed hierarchy as in Ward clustering but the dendrogram gives moreover simple decision rules for each cluster of the hierarchy. This method is designed for numerical, categorical and mixed data (mixture of numerical and numerical variables) and is implemented is the divclust R package.