Belief Fusion consists of merging beliefs about a domain of interest from multiple separate sources. No single belief fusion method is adequate for all categories of situations, hence the challenge is to determine which belief fusion method is the most appropriate for a given situation. The conclusion to be drawn from this discussion is that the analyst must first understand the dynamics of the situation at hand in order to find the best fusion method for analysing it. The aim of this article is first to demonstrate that there are appropriate situations to use belief fusion, and that different mathematical fusion operators are required for the different situations. Secondly we propose criteria than can be applied to identify the various categories of fusion situations, and describe specific belief fusion operators that are suitable for modeling the fusion situations in each category