Insulin-dependent diabetes mellitus (IDDM) results from the destruction of pancreatic insulin-secreting cells by a T-cell-mediated autoimmune reaction. Distinct types of T helper cells (TH1 and TH2) have been characterised based on their cytokine secretion profiles following activation. Evidence from animal models favours the hypothesis that autoimmune diabetes is a TH1 response. However, there is no clear indication that a primary imbalance between protective TH2 and deleterious TH1 cells at early stages can trigger the autoimmune process. Protective CD4 + cells detected in nondiabetic young non-obese diabetic mice have not been shown to work through TH2 cytokines. In humans, there is little evidence that IDDM results from a TH1 response. Indeed, efficient experimental systems are lacking in humans to study the regulation of the autoimmune response in vitro. Interestingly, several immunotherapy strategies have aimed at inducing a TH2 response, even though TH2 cells have not been implicated in spontaneous disease development. However, recent ongoing trials in humans using oral administration of insulin to prevent diabetes are based on a protective mechanism which seems to depend essentially on transforming growth factor-beta. This cytokine is not dependent on TH1/TH2 dichotomy. Thus, although several attempts have been made to induce a TH1/TH2 switch to obtain a protective effect, a different and more complex mechanism probably (and paradoxically) accounts for the oral protection actually tested in animal models and humans.