In this article we establish a classification and we discuss the diffe
rent ''problematiques'' that may be in question in a certain stage of
a decision-aid study. We enlarge the trilogy ''choice, sorting and ran
king'', introduced by Bernard Roy, reinforcing the importance of this
subject in the methodological context of decision-aid. First we distin
guish decisional problematique from technical or decision-aiding probl
ematique. Afterwards, we reflect oil three questions: Do we want a rel
ative comparison or an absolute evaluation of the actions?, to make a
choice or to rank the actions?, to accept or to reject actions?