It remains a challenge to differentiate human induced pluripotent stem cells (iPSCs) or embryonic stem (ES) cells to Purkinje cells. In this study, we derived iPSCs from human fibroblasts and directed the specification of iPSCs first to Purkinje progenitors, by adding Fgf2 and insulin to the embryoid bodies (EBs) in a time-sensitive manner, which activates the endogenous production of Wnt1 and Fgf8 from EBs that further patterned the cells towards a midbrain-hindbrain-boundary tissue identity. Neph3-positive human Purkinje progenitors were sorted out by using flow cytometry and cultured either alone or with granule cell precursors, in a 2-dimensional or 3-dimensional environment. However, Purkinje progenitors failed to mature further under above conditions. By co-culturing human Purkinje progenitors with rat cerebellar slices, we observed mature Purkinje-like cells with right morphology and marker expression patterns, which yet showed no appropriate membrane properties. Co-culture with human fetal cerebellar slices drove the progenitors to not only morphologically correct but also electrophysiologically functional Purkinje neurons. Neph3-posotive human cells could also survive transplantation into the cerebellum of newborn immunodeficient mice and differentiate to L7- and Calbindin-positive neurons. Obtaining mature human Purkinje cells in vitro has significant implications in studying the mechanisms of spinocerebellar ataxias and other cerebellar diseases.