Immune tolerance to self-antigens is physiological. Given a repertoire of self-reactive, potentially pathogenic lymphocytes, therapeutic options to diminish autoimmune disease risk include deletion, reduced activation or increased regulation of self-reactive lymphocytes by means that mimic or promote physiological mechanisms of immunity. Vaccination with self-antigen to promote self-antigen-specific tolerance, 'negative vaccination', may represent the most specific and potentially safest means of averting autoimmune disease. This strategy is therapeutically effective in inbred rodent models but its translation in humans has failed to meet expectations. This failure can be attributed to the use of suboptimal dosage regimens in end-stage disease, as well as other factors. This review focuses on vaccination against self-antigen in type 1 diabetes, an autoimmune disease unique in that individuals at risk can be identified years before clinical presentation. Moreover, the spontaneously diabetic non-obese diabetic mouse, which mimics human type 1 diabetes in many ways, has provided 'proof-of-concept' for negative vaccination. Recent trials of a nasal insulin vaccine in humans at risk of type 1 diabetes provide evidence of tolerance induction as a basis for clinical efficacy.