As yet, few molecular markers are available that are likely to be useful in predicting the metastatic potential of prostate cancer cancer cells. The need for such "progression markers" is indicated by the expectation that more localized cancers (ie stage A-2, B-1,2) will be clinically diagnosed in the near future, owing to improvements in diagnostic techniques (eg transrectal ultrasound) and the screening of population groups at risk (males over 50 years old). Few model systems are available for such studies. Animal models are unsuitable for the isolation of monoclonal antibodies that detect epitopes associated with the progression of prostate cancer. Since few cell lines are available, an approach using primary cancer tissue should be undertaken. For the differential hybridization approach described here, the choice of species presents no difficulty, since many DNA sequences are conserved between species. However, no model system fully represents the human situation. Hence, differential hybridization studies using primary prostate cancer tissue need to be considered. Moreover, the group of genes/proteins with potential relevance for cancer progression (eg oncogenes, tumour suppressor genes, genes encoding cell adhesion molecules or growth factors) has not been studied extensively in prostate cancer. Because of the intrinsic heterogeneity of prostate tumours, the use of pathologically defined tissue sections is imperative for a reliable study. This could be achieved either by in situ techniques (whereby tissue morphology is conserved) or by step sectioning. The difficulties associated with the small amount of material obtained from such sections can be overcome by the use of techniques based on the polymerase chain reaction. Taking these considerations into account, a systematic screening of prostate cancer with probes representing the above mentioned genes should be undertaken.