Diabetes mellitus is characterized by the loss of insulin-producing beta cells. While conventional treatment results in severe long-term complications, cell replacement therapy is a promising approach for the cure of this disease. However, its application is severally limited by the shortage of donor tissue. Hence, great research efforts concentrate on the development of an abundant cell source of functional beta-like cells, by pursuing three main strategies: Expansion of human donor beta cells in vitro, reprogramming of other cell types, and directed differentiation of pluripotent stem cells, both embryonic and patient-derived. The goal of all these approaches has been the generation of cells with properties that closely resemble the beta-cell phenotype, in particular production and storage of adequate amounts of mature insulin, and its regulated release in response to physiological signals. Here we review recent progress in all three approaches and discuss their advantages as well as remaining challenges.