Under normal conditions, the regeneration of mouse beta cells is mainly dependent on their own duplication. Although there is evidence that pancreatic progenitor cells exist around duct, whether non-beta cells in the islet could also potentially contribute to beta cell regeneration in vivo is still controversial. Here, we developed a novel transgenic mouse model to study the pancreatic beta cell regeneration, which could specifically inhibit beta cell proliferation by overexpressing p21(cip) in beta cells via regulation of the Tet-on system. We discovered that p21 overexpression could inhibit beta-cell duplication in the transgenic mice and these mice would gradually suffer from hyperglycemia. Importantly, the recovery efficiency of the p21-overexpressing mice from streptozotocin-induced diabetes was significantly higher than control mice, which is embodied by better physiological quality and earlier emergence of insulin expressing cells. Furthermore, in the islets of these streptozotocin-treated transgenic mice, we found a large population of proliferating cells which expressed pancreatic duodenal homeobox 1 (PDX1) but not markers of terminally differentiated cells. Transcription factors characteristic of early pancreatic development, such as Nkx2.2 and NeuroD1, and pancreatic progenitor markers, such as Ngn3 and c-Met, could also be detected in these islets. Thus, our work showed for the first time that when beta cell self-duplication is repressed by p21 overexpression, the markers for embryonic pancreatic progenitor cells could be detected in islets, which might contribute to the recovery of these transgenic mice from streptozotocin-induced diabetes. These discoveries could be important for exploring new diabetes therapies that directly promote the regeneration of pancreatic progenitors to differentiate into islet beta cells in vivo.