How trivial are reliable box-plane intersections?

Abstract Three completely different algorithms for box–plane intersection testing are considered. They are compared with respect to worst case analysis, average performance and overhead. It turns out that each of the algorithms has different advantages and it is difficult to select one of them. Particular new aspects include the observation that one only needs two vertices on a particular diagonal of the box to determine the intersection with the plane and that the algorithm ESSA is used to resolve cases left indeterminate by the main algorithm because of rounding errors.