Tumors express tumor-associated antigens (TAA) and thus should be the object of immune attack. Nonetheless, spontaneous clearance of established tumors is rare. Much work has demonstrated that tumors have numerous strategies either to prevent presentation of TAA, or to prevent TAA presentation in the context of T-cell co-signaling molecules. Thus, it was thought that lack of TAA-specific immunity was largely a passive process: tumors simply did not present enough TAA, or antigen-presenting cells did not have sufficient stimulatory capacity. On this basis, attempts were made to bolster TAA-specific immunity by using optimal antigen-presenting cells or by growing TAA-specific effector T cells ex vivo followed by adoptive transfer. These approaches met with some success in mouse models of human tumors, and showed some early clinical efficacy in human trials, although long-term efficacy remains to be established, and logistical problems are considerable. These studies established the concept that experimentally induced TAA-specific immunity is a rational and potentially efficacious means to treat cancer, including ovarian cancer. Nonetheless, recent work demonstrates that lack of naturally induced TAA-specific immunity is not simply a passive process. We discuss recent data clearly demonstrating that 'tumors actively prevent induction of TAA-specific immunity through induction of TAA-specific tolerance'. This tolerance is mediated in part by regulatory T cells (Tregs). Means to revert these tolerizing conditions represent a novel anticancer therapeutic stratagem. We discuss Tregs in this regard in human ovarian cancer and present evidence that depleting Treg in human cancer, including ovarian cancer, using denileukin diftitox (Ontak), improves immunity and may be therapeutic.