Progressive encephalopathy with oedema, hypsarrhythmia, and optic atrophy (PEHO) syndrome is an early childhood onset, severe autosomal recessive encephalopathy characterized by extreme cerebellar atrophy due to almost total granule neuron loss. By combining homozygosity mapping in Finnish families with Sanger sequencing of positional candidate genes and with exome sequencing a homozygous missense substitution of leucine for serine at codon 31 in ZNHIT3 was identified as the primary cause of PEHO syndrome. ZNHIT3 encodes a nuclear zinc finger protein previously implicated in transcriptional regulation and in small nucleolar ribonucleoprotein particle assembly and thus possibly to pre-ribosomal RNA processing. The identified mutation affects a highly conserved amino acid residue in the zinc finger domain of ZNHIT3. Both knockdown and genome editing of znhit3 in zebrafish embryos recapitulate the patients' cerebellar defects, microcephaly and oedema. These phenotypes are rescued by wild-type, but not mutant human ZNHIT3 mRNA, suggesting that the patient missense substitution causes disease through a loss-of-function mechanism. Transfection of cell lines with ZNHIT3 expression vectors showed that the PEHO syndrome mutant protein is unstable. Immunohistochemical analysis of mouse cerebellar tissue demonstrated ZNHIT3 to be expressed in proliferating granule cell precursors, in proliferating and post-mitotic granule cells, and in Purkinje cells. Knockdown of Znhit3 in cultured mouse granule neurons and ex vivo cerebellar slices indicate that ZNHIT3 is indispensable for granule neuron survival and migration, consistent with the zebrafish findings and patient neuropathology. These results suggest that loss-of-function of a nuclear regulator protein underlies PEHO syndrome and imply that establishment of its spatiotemporal interaction targets will be the basis for developing therapeutic approaches and for improved understanding of cerebellar development.