We have generated anti-HER2 (ErbB2) immunoliposomes (ILs), consisting of long circulating liposomes linked to anti-HER2 monoclonal antibody (MAb) fragments, to provide targeted drug delivery to HER2-overexpressing cells. Immunoliposomes were constructed using a modular strategy in which components were optimized for internalization and intracellular drug delivery. Parameters included choice of antibody construct, antibody density, antibody conjugation procedure, and choice of liposome construct. Anti-HER2 immunoliposomes bound efficiently to and internalized in HER2-overexpressing cells in vitro as determined by fluorescence microscopy, electron microscopy, and quantitative analysis of fluorescent probe delivery. Delivery via ILs in HER2-overexpressing cells yielded drug uptake that was up to 700-fold greater than with non-targeted sterically stabilized liposomes. In vivo, anti-HER2 ILs showed extremely long circulation as stable constructs in normal adult rats after a single i.v. dose, with pharmacokinetics that were indistinguishable from sterically stabilized liposomes. Repeat administrations revealed no increase in clearance, further confirming that ILs retain the long circulation and non-immunogenicity of sterically stabilized liposomes. In five different HER2-overexpressing xenograft models, anti-HER2 ILs loaded with doxorubicin (dox) showed potent anticancer activity, including tumor inhibition, regressions, and cures (pathologic complete responses). ILs were significantly superior vs. all other treatment conditions tested: free dox, liposomal dox, free MAb (trastuzumab), and combinations of dox+MAb or liposomal dox+MAb. For example, ILs produced significantly superior antitumor effects vs. non-targeted liposomes (P values from <0.0001 to 0.04 in eight separate experiments). In a non-HER2-overexpressing xenograft model (MCF7), ILs and non-targeted liposomal dox produced equivalent antitumor effects. Detailed studies of tumor localization indicated a novel mechanism of drug delivery for anti-HER2 ILs. Immunotargeting did not increase tumor tissue levels of ILs vs. liposomes, as both achieved very high tumor localization (7.0-8.5% of injected dose/g tissue) in xenograft tumors. However, histologic studies using colloidal-gold labeled ILs demonstrated efficient intracellular delivery in tumor cells, while non-targeted liposomes accumulated within stroma, either extracellularly or within macrophages. In the MCF7 xenograft model lacking HER2-overexpression, no difference in tumor cell uptake was seen, with both ILs and non-targeted liposomes accumulating within stroma. Thus, anti-HER2 ILs, but not non-targeted liposomes, achieve intracellular drug delivery via receptor-mediated endocytosis, and this mechanism is associated with superior antitumor activity. Based on these results, anti-HER2 immunoliposomes have been developed toward clinical trials. Reengineering of construct design for clinical use has been achieved, including: new anti-HER2 scFv F5 generated by screening of a phage antibody library for internalizing anti-HER2 phage antibodies; modifications of the scFv expression construct to support large scale production and clinical use; and development of methods for large-scale conjugation of antibody fragments with liposomes. We developed a scalable two-step protocol for linkage of scFv to preformed and drug-loaded liposomes. Our final, optimized anti-HER2 ILs-dox construct consists of F5 conjugated to derivatized PEG-PE linker and incorporated into commercially available liposomal doxorubicin (Doxil). Finally, further studies of the mechanism of action of anti-HER2 ILs-dox suggest that this strategy may provide optimal delivery of anthracycline-based chemotherapy to HER2-overexpressing cancer cells in the clinic, while circumventing the cardiotoxicity associated with trastuzumab+anthracycline. We conclude that anti-HER2 immunoliposomes represent a promising technology for tumor-targeted drug delivery, and that this strategy may also be applicable to other receptor targets and/or using other delivered agents.