Chemotherapy, radiation, and surgery are the conventional treatment modalities for cancer. The success achieved with these approaches has been limited due to several factors like chemoresistance to drugs, non-specificity leading to peripheral toxicity, and non-resectable tumors. To combat these problems, the concept of targeted therapy using immunotoxins was developed. Immunotoxins are chimeric proteins with a cell-selective ligand chemically linked or genetically fused to a toxin moiety and can target cancer cells overexpressing tumor-associated antigens, membrane receptors, or carbohydrate antigens. Ligands for these receptors or monoclonal antibodies or single chain variable fragments directed against these antigens are fused with bacterial or plant toxins and are made use of as immunotoxins. Pseudomonas exotoxin, anthrax toxin, and diphtheria toxin are the commonly used bacterial toxins. Ricin, saporin, gelonin, and poke weed antiviral protein are the plant toxins utilized in immunotoxin constructs. Several such fusion proteins are in clinical trials, and denileukin difitox is a FDA-approved fusion protein. In spite of the promise shown by bacterial- and plant toxin-based chimeric proteins, their clinical application is hampered by several factors like immunogenicity of the toxin moiety and non-specific toxicity leading to vascular leak syndrome. In order to overcome these problems, a novel generation of immunotoxins in which the cytotoxic moiety is an endogenous protein of human origin like proapoptotic protein or RNase has been developed. This review summarizes the advances in this new class of fusion protein and the future directions to be explored. © 2009 Japanese Cancer Association.