GLIS3 mutations are associated with type 1, type 2, and neonatal diabetes, reflecting a key function for this gene in pancreatic β-cell biology. Previous attempts to recapitulate disease-relevant phenotypes in GLIS3-/- β-like cells have been unsuccessful. Here, we develop a "minimal component" protocol to generate late-stage pancreatic progenitors (PP2) that differentiate to mono-hormonal glucose-responding β-like (PP2-β) cells. Using this differentiation platform, we discover that GLIS3-/- hESCs show impaired differentiation, with significant death of PP2 and PP2-β cells, without impacting the total endocrine pool. Furthermore, we perform a high-content chemical screen and identify a drug candidate that rescues mutant GLIS3-associated β-cell death both in vitro and in vivo. Finally, we discovered that loss of GLIS3 causes β-cell death, by activating the TGFβ pathway. This study establishes an optimized directed differentiation protocol for modeling human β-cell disease and identifies a drug candidate for treating a broad range of GLIS3-associated diabetic patients.