Decision models are often based on certain assumptions as to their val
idity. Relevant assumptions may include value-based assumptions, such
as limitations on the range or values of some input variables or exoge
nous factors, as well as assumptions about model structure (e.g., Line
arity). In a model base consisting of many models, there may be severa
l models (or collections of models) that can be used to solve a partic
ular problem. We may wish to know what the applicable models are, what
assumptions are associated with these models, and whether a given set
of assumptions is necessary and/or sufficient for solving the problem
. We describe an analytical approach, based on a graph-theoretic const
ruct called a metagraph, and show how it can be used to represent and
analyze assumptions in model bases.