Résumé
Recently, we have introduced an uncertainty representation generalising imprecise cumulative distributions to any (pre-)ordered space as well as possibility distributions: generalised p-boxes. This representation has many attractive features, as it remains quite simple while having an interesting interpretation in terms of lower and upper confidence bounds over nested sets. However, the merits of this representation in various uncertainty processing tasks still have to be evaluated. This is the topic of this paper, in which the handling of information modelled by generalised p-boxes is studied, from the point of view of elicitation, propagation, conditioning and fusion.