The proofs of quantum nonlocality due to Greenberger, Horne and Zeilinger and due to Hardy are qualitatively different from that of Bell insofar as they rely only on a consideration of whether events are possible or impossible, rather than relying on specific experimental probabilities. We consider the scenario of a bipartite nonlocality experiment, in which two separated experimenters each have access to some measurements they can perform on a system. In a physical theory, some outcomes of this experiment will be labelled possible, others impossible, and an assignment of the values 0 (impossible) and 1 (possible) to these different outcomes forms a table of possibilities. Here, we consider the computational task of determining whether or not a given table of possibilities constitutes a departure from possibilistic local realism. By considering the case in which one party has access to measurements with two outcomes and the other three, it is possible to see at exactly which point this task becomes computationally difficult.