The probabilities in Bayes nets are themselves uncertain, often based on expert judgement or limited, collateral data. It does not seem sensible to specify a distribution for this uncertainty because it requires information about shapes and moments that usually is unavailable. Instead, the uncertainties in a Bayes net may be represented as sets of intervals. Information gap theory provides a platform for interpreting the uncertainties relative to a decision. This approach is illustrated with an example of the management of arboreal mammals in southern NSW forests.