Mining data warehouses is still an open problem as few approaches really take into account the specifities of this framework (e.g. multidimensionnality, hierarchies, historized data). Multidimensional sequential patterns have been studied. However, they do not provide any way to handle hierarchies. In this paper, we propose an original method of extraction of sequential patterns taking into account the hierarchies. This method extracts more accurate knowledge and extends our preceding approach \m. We define the concepts related to our problems as well as the associated algorithms. The experiments which we carried out show the interest of our proposal.