Gene therapy for breast cancer initially involves local or systemic delivery. Local delivery may be intrapleural or via direct injection to lesions. However, systemic delivery remains the greatest challenge with targeting, although methods using antibodies or growth factor receptor ligands have been demonstrated in preclinical models. This review focuses on the next step of using tissue-specific promoters such as Muc-1, CEA, PSA, HER-2, Myc, L-plastin and secretory leukoproteinase inhibitor promoters. All of these have demonstrated differential upregulation in breast cancer and additional specificity may be obtained by using physiological stimuli that are more frequently expressed in cancers, such as glucose regulated promoters and hypoxia response elements or radiation inducible elements. Amongst the later are the EGR-1, p21 and tissue type plaminogen activator promoters. Potential therapy genes include the prodrug activation system 5-fluorocytosine and other analogues of antimetabolites, but all of these need gap junctions to transfer the phosphorylated metabolites. Other approaches involving more freely diffusible products include cyclophosphamide, ifosfamide and thymidine phosphorylase to activate 5-deoxy-5-fluoruridine to fluorouracil. The bystander effect is important both for cell killing and for immunological and antivascular effects. Breast cancer is one type of tumour where a major clinical research effort is underway using local delivery methods. For prodrug activation systems, the use of human enzymes is desirable to prevent an immunological response that would eventually eliminate cells producing the prodrug activation system. The use of alkylating agents has an advantage over antimetabolites in that they are cytotoxic to cycling and noncycling cells, and the cytotoxic products can diffuse across cell membranes without the need for gap junctions. They also have a much steeper dose response curve than antimetabolites.