The rationale and limitations of discriminating between cases of schizophrenia with and without a family history are reviewed. It is concluded from the evidence available that, by identifying subgroups of greater aetiological homogeneity, the strategy can be a useful starting point for research into likely causes.