Germline loss-of-function variation in TNFAIP3, encoding A20, has been implicated in a wide variety of autoinflammatory and autoimmune conditions, with acquired somatic missense mutations linked to cancer progression. Furthermore, human sequence data reveals that the A20 locus contains ~ 400 non-synonymous coding variants, which are largely uncharacterised. The growing number of A20 coding variants with unknown function, but potential clinical impact, poses a challenge to traditional mouse-based approaches. Here we report the development of a novel functional genomics approach that utilizes a new A20-deficient zebrafish (Danio rerio) model to investigate the impact of TNFAIP3 genetic variants in vivo. A20-deficient zebrafish are hyper-responsive to microbial immune activation and exhibit spontaneous early lethality. Ectopic addition of human A20 rescued A20-null zebrafish from lethality, while missense mutations at two conserved A20 residues, S381A and C243Y, reversed this protective effect. Ser381 represents a phosphorylation site important for enhancing A20 activity that is abrogated by its mutation to alanine, or by a causal C243Y mutation that triggers human autoimmune disease. These data reveal an evolutionarily conserved role for TNFAIP3 in limiting inflammation in the vertebrate linage and show how this function is controlled by phosphorylation. They also demonstrate how a zebrafish functional genomics pipeline can be utilized to investigate the in vivo significance of medically relevant human TNFAIP3 gene variants.