As oral administration of insulin reduces the incidence of diabetes in NOD mice, and to achieve a better approximation of oral insulin trials being developed for human studies which will use human insulin, we attempted to determine the preventive efficacy of oral administration of human insulin rather than resorting to the animal insulins used in previous studies. As the strength of prevention obtained by oral insulin has not been adequately demonstrated, we determined whether the protection persisted after the oral treatment was discontinued and whether it was resistant to a diabetogenic injection of cyclophosphamide (CY). We also determined whether the effect of insulin could be increased by oral administration of lipopolysaccharide from Escherichia coli (LPS) or another immunostimulant (glycoprotein extracts from Klebsiella pneumoniae, GEKP) which may be more feasible for human application. Female NOD mice were fed once a week (from 35 to 300 days of age) with insulin, LPS, GEKP, insulin plus LPS, insulin plus GEKP, or PBS. A decreased incidence of diabetes were observed in animals fed human insulin (p < 0.01 incidence of diabetes at 300 days of age: 31% in mice fed with insulin and 65% in those fed PBS). Prevention by insulin was not enhanced by oral LPS or GEKP. Yet unexpectedly, mice fed with LPS alone or GEKP alone displayed decreases in diabetes incidence (p < 0.01). The severity of insulitis was reduced in animals fed insulin, LPS, GEKP or combinations of insulin and either immunostimulant (p < 0.02). Although the oral treatments were stopped at 300 days of age, the incidence of diabetes at 360 days remained lower in mice previously fed insulin, LPS, GEKP or combinations of insulin and either immunostimulant (p < 0.01). In mice previously fed PBS, CY injection (60 days after withdrawal of the oral treatment) led to a final incidence of diabetes of 90% (sum of the incidence during the initial 360 days and the further CY-induced incidence). Previous feedings with insulin, LPS, GEKP or combinations of insulin and either immunostimulant did not protect against CY-induced diabetes since incidences reached the final control incidence. T splenocytes from animals fed insulin, LPS, or GEKP, similarly reduced the capacity of T cells from diabetic mice to transfer the disease (p < 0.01). It is concluded that oral treatment with human insulin to be used in human trials reduces the incidence of diabetes in NOD mice. Equivalent preventive efficacy was obtained through feedings with LPS or GEKP (even though no cumulative efficiency was observed with insulin). The latter results suggest that it would be advisable to evaluate the efficiency of oral bacterial antigens for the prevention of human Type 1 diabetes. The protection afforded by oral treatments with insulin or bacterial antigens may be attributed to cellular suppression, persists for some time after treatments are stopped, but is not resistant to major immune stimulation such as injection of CY.