The antitumour effects of interferons in animals and humans are well known. Despite the fact, however, that the 3 types of human interferon, leukocyte alpha-interferon, fibroblast beta-interferon and immune gamma-interferon are available in large amounts through recombinant DNA technology, the practical applicability of interferon therapy in cancer is still not clear. An initial approach t o this problem is t o determine the mechanism of action of interferons and t o find out why, in certain circumstances, they are inactive. There are various ways in which interferon may control tumoursi.e. antiviral action, inhibition of cell growth, stimulation of cell differentiation, changes in cells modulating the susceptibility t o immune rejection, or effects on the host immune systems (natural killer system and cytotoxic proteins). The implications of these data in the use of interferon in cancer therapy need t o be evaluated. Both alpha-and beta-interferons may have beneficial effects on growth inhibition and differentiation, but gamma-interferon is probably more effective in boosting the immune recognition and rejection of tumour cells. A combination of alpha-and gamma-interferon may give the best results in vivo, since they often act synergistically in vitro. The sensitivity of individual tumour cells t o the various types of interferon needs t o be evaluated by measurement of oncogenes mRNA inhibition, Go/G, arrest and increase in various H-La antigens. Finally, the aim of any treatment (antiviral action, tumour regression, prevention of metastasis, decreased tumour growth and increased cell differentiation) should be an important consideration in whether interferon therapy is chosen. A major problem remains in understanding why only a small proportion of patients usually show an objective response t o interferon. This could be due t o individual differences in their sensitivity t o interferon, or because patients benefit from interferon therapy only if they are deficient in interferon or i f i t s synthesis is inhibited,