β-Cell replacement represents an attractive prospect for diabetes therapy. Although much hope has been placed on derivation of insulin-producing cells from human pluripotent stem cells, this approach continues to face considerable challenges. Cells from adult human tissues, with both stem/progenitor and mature phenotypes, offer a possible alternative. This review summarizes recent progress in two major strategies based on this cell source, . ex vivo expansion of human islet β cells and conversion of non-β cells into insulin-producing cells by nuclear reprogramming, and examines the obstacles that remain to be overcome for bringing these strategies closer to clinical application in diabetes therapy.