In animal systems, complete and permanent eradication of tumours can be achieved by adoptive transfer of MHC-restricted T cells, combined with IL2. In certain types of human cancer (melanoma and perhaps renal cell carcinoma), tumour-specific T cells are probably the therapeutically most active cells among LAK or TIL cells. To prove these points, it is necessary to conduct trials with cloned tumour-specific T cells. Other potentially immunogenic tumors are cervical carcinoma, associated with human papilloma virus, and Burkitt's lymphoma, associated with Epstein-Barr virus. Most other human tumours, caused by subtle mutations in proto-oncogenes, are likely to be poorly or non-immunogenic. It is worthwhile trying to overcome this by vaccination with IL2 or IFN gamma-producing tumour cells or by deliberate vaccination with desirable targets for tumour-specific CTL such as the products of point-mutated oncogenes, including ras (Jung and Schleusener, 1991) and p53 (Rodriguez et al., 1990; Halevy et al., 1990), provided the relevant peptides are processed and bound to MHC class I molecules. Other potential targets are breakpoint peptides of translocated oncogene products such as bcr/abl (Van Denderen et al., 1990). In viral systems, it has already been established that peptide vaccination for protective CTL induction is feasible (Aichele et al., 1989; Schulz et al., 1991; Kast et al., 1991).