Stem cells, derived from either embryonic or adult tissues, are considered to be potential sources of insulin-secreting cells to be transplanted into type 1 and advanced stages of type 2 diabetic patients. Many laboratories have considered this possibility, resulting in a large amount of published protocols, with a wide degree of complexity among them. Our group was the first to report that it was possible to obtain insulin-secreting cells from mouse embryonic stem cells, proving the feasibility of this new challenge. The same observation was immediately reported using human embryonic stem cells. However, the resulting cell product was not properly characterised, affecting the reproducibility of the protocol by other groups. A more elaborated protocol was developed by Lumelsky and co-workers, demonstrating that neuroectodermal cells could be an alternative source for insulin-producing cells. However, the resulting cells of this protocol produced low amounts of the hormone. This aimed other groups to perform key changes in order to improve the insulin content of the resulting cells. Recently, Baetge's group has published a new protocol based on the knowledge accumulated in pancreatic development. In this protocol, human embryonic stem cells were differentiated into islet-like structures through a five step protocol, emulating the key steps during embryonic development of the endocrine pancreas. The final cell product, however, seemed to be in an immature state, thus further improvement is required. Despite this drawback, the protocol represents the culmination of work performed by different groups and offers new research challenges for the investigators in this exciting field. Concerning adult stem cells, the possibility of identifying pancreatic precursors or of reprogramming extrapancreatic derived cells are key possibilities that may circumvent the problems that appear when using embryonic stem cells, such as immune rejection and tumour formation.