Gene therapy was initially thought of as a means to correct single gene defects in hereditary disease. In the meantime, cancer has become by far the most important indication for gene therapy in clinical trials. In the foreseeable future, the best way to achieve reasonable intratumoral concentrations of a transgene with available vectors is direct intratumoral injection with or without the aid of various techniques such as endoscopy or CT-guidance. At present, viral and non-viral methods of gene transfer are used either in vivo or ex vivo/in vitro. The most important viral vectors currently in use in clinical trials comprise retroviruses, adenoviruses, adeno-associated viruses, and herpes viruses. None of the available vectors satisfies all the criteria of an ideal gene therapeutic system, and vectors with only minimal residues of their parent viruses ("gutless vectors") as well as completely "synthetic viral vectors" will gain more and more importance in the future. Non-viral gene therapy methods include liposomes, injection of vector-free DNA ("naked DNA"), protein-DNA complexes, delivery by "gene gun," calcium-phosphate precipitation, electroporation, and intracellular microinjection of DNA. The first clinical trial of gene therapy for cancer was performed in 1991 in patients with melanoma, and since then more than 5000 patients have been treated worldwide in more than 400 clinical protocols. With the exception of a case of fatal toxicity in a young man with hereditary liver disease treated intrahepatically with high doses of adenovirus, side effects have been rare and usually mild in all these studies and expression of the transgene could be demonstrated in patients in vivo. However, despite anecdotal reports of therapeutic responses in some patients, unequivocal proof of clinical efficacy is still lacking for most of the varied approaches to gene therapy in humans. As well as our only fragmentary understanding of the molecular pathophysiology of many diseases, the principal reason for the present lack of clinical success of gene therapy is the very low transduction and expression efficiency in vivo of available vectors. Despite the complexities of gene therapy for cancer, the numerous different approaches can be subdivided into three basic concepts: (1) strengthening of the immune response against a tumour, (2) repair of cell cycle defects caused by losses of tumour suppressor genes or inappropriate activation of oncogenes, and (3) suicide gene strategies. In addition, the importance of gene marker studies and gene therapeutic protection of normal tissue are briefly covered in this review.