Human Vγ2Vδ2 T cells monitor isoprenoid metabolism by recognizing foreign (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate (HMBPP), a metabolite in the 2-C-methyl-D-erythritol-4-phosphate pathway used by most eubacteria and apicomplexan parasites, and self isopentenyl pyrophosphate, a metabolite in the mevalonate pathway used by humans. Whereas microbial infections elicit prolonged expansion of memory Vγ2Vδ2 T cells, immunization with prenyl pyrophosphates or aminobisphosphonates elicit short-term Vγ2Vδ2 expansion with rapid anergy and deletion upon subsequent immunizations. We hypothesized that a live, attenuated bacterial vaccine that overproduces HMBPP would elicit long-lasting Vγ2Vδ2 T cell immunity by mimicking a natural infection. Therefore, we metabolically engineered the avirulent aroA(-) Salmonella enterica serovar Typhimurium SL7207 strain by deleting the gene for LytB (the downstream enzyme from HMBPP) and functionally complementing for this loss with genes encoding mevalonate pathway enzymes. LytB(-) Salmonella SL7207 had high HMBPP levels, infected human cells as efficiently as did the wild-type bacteria, and stimulated large ex vivo expansions of Vγ2Vδ2 T cells from human donors. Importantly, vaccination of a rhesus monkey with live lytB(-) Salmonella SL7207 stimulated a prolonged expansion of Vγ2Vδ2 T cells without significant side effects or anergy induction. These studies provide proof-of-principle that metabolic engineering can be used to derive live bacterial vaccines that boost Vγ2Vδ2 T cell immunity. Similar engineering of metabolic pathways to produce lipid Ags or B vitamin metabolite Ags could be used to derive live bacterial vaccine for other unconventional T cells that recognize nonpeptide Ags.