Stem cells might provide unlimited supply of transplantable cells for β-cell replacement therapy in diabetes. The human limbus is a highly specialized region hosting a well-recognized population of epithelial stem cells, which sustain the continuous renewal of the cornea, and the recently identified stromal fibroblast-like stem cells (f-LSCs), with apparent broader plasticity. However, the lack of specific molecular markers for the identification of the multipotent limbal subpopulation has so far limited the investigation of their differentiation potential. In this study we show that the human limbus contains uncommitted cells that could be potentially harnessed for the treatment of diabetes. Fourteen limbal biopsies were obtained from patients undergoing surgery for ocular diseases not involving the conjunctiva or corneal surface. We identified a subpopulation of f-LSCs characterized by robust proliferative capacity, expressing several pluripotent stem cell markers and exhibiting self-renewal ability. We then demonstrated the potential of f-LSCs to differentiate in vitro into functional insulin-secreting cells by developing a four-step differentiation protocol that efficiently directed f-LSCs towards the pancreatic endocrine cell fate. The expression of specific endodermal, pancreatic, islet, and β-cell markers, as well as functional properties of f-LSC-derived insulin-producing cells, were evaluated during differentiation. With our stage-specific approach, up to 77% of f-LSCs eventually differentiated into cells expressing insulin (also assessed as C-peptide) and exhibited phenotypic features of mature β-cells, such as expression of critical transcription factors and presence of secretory granules. Although insulin content was about 160-fold lower than what observed in adult islets, differentiated cells processed ∼98% of their proinsulin content, similar to mature β-cells. Moreover, they responded in vitro in a regulated manner to multiple secretory stimuli, including glucose. In conclusion, f-LSCs represent a possible relevant source of autologous, transplantable, insulin-producing cells that could be tested for the reversal of diabetes.