An ideal therapeutic for cancer would be one that selectively targets to tumor cells, is nontoxic to normal cells, and that could be systemically delivered, thereby reaching metastases as well as primary tumor. Immunoliposomes directed by monoclonal antibody or its fragments are promising vehicles for tumor-targeted drug delivery. However, there is currently very limited data on gene delivery using these vehicles. We have recently described a cationic immunoliposome system directed by a lipid-tagged, single-chain antibody Fv fragment (scFv) against the human transferrin receptor (TfR) that shows promising efficacy for systemic p53 tumor suppressor gene therapy in a human breast cancer metastasis model. However, the extremely low yield of this lipid-tagged scFv limited further downstream development and studies. Here we report a different expression strategy for the anti-TfR scFv, which produces high levels of protein without any tags, and a different approach for complexing the targeting scFv to the liposomes. This approach entails covalently conjugating the scFv to the liposome via a cysteine at the 3'-end of the protein and a maleimide group on the liposome. Our results show that this conjugation does not impair the immunological activity or targeting ability of the scFv. The scFv-cys targets the cationic liposome-DNA complex (lipoplex) to tumor cells and enhances the transfection efficiencies both in vitro and in vivo in a variety of human tumor models. This scFv-immunoliposome can deliver the complexed gene systemically to tumors in vivo, where it is efficiently expressed. In comparison with the whole antibody or transferrin molecule itself, the scFv has a much smaller size for better penetration into solid tumors. It is also a recombinant protein rather than a blood product; thus, large scale production and strict quality control are feasible. This new approach provides a promising system for tumor-targeted gene delivery that may have potential for systemic gene therapy of various human cancers.