Sets of desirable gambles are a very general type of model for uncertainty. They
are a bit more general than closed convex sets of probabilities, but there is
nevertheless a close correspondence between both types. Due to the geometric
flavor distinguishing them from models based directly on probabilities, they
provide a fresh way of looking at modeling uncertainty that can be inspiring
even if one wishes to keep models with a probabilistic flavor central. I will
give an introduction to the theory of sets of desirable gambles and make the
relationship with probabilisticly-flavored  models for uncertainty clear.