The study of integrin receptor expression and function in carcinomas will undoubtedly increase our understanding of the malignant process and clarify the relative importance of the numerous alterations that are identified. Whether this will lead to direct clinical applications is not clear at present, but it will certainly alter our approach to the evaluation of cancer treatment. The more immediate search for prognostic indicators and metastatic site preference in tumours expressing a particular integrin profile is partly hampered by the small numbers of cases studied in most series, where observed trends in tumour subsets barely achieve significance. Perhaps the next phase should be to evaluate promising approaches in large multicentre studies to establish the relative importance of these trends in tumour prognosis and response to treatment, preferably by developing reagents that will permit the demonstration of integrins in routinely processed archival material from human tumour specimens. Future developments in therapy may depend on the knowledge that may emerge from such work.