Asparaginase, an enzyme therapeutic that deaminates asparagine to produce aspartate and ammonia, is among the most effective treatments for several cancers. However, the molecular determinants governing asparaginase response and resistance remain poorly defined, limiting its therapeutic potential. We previously demonstrated that asparaginase has potent activity in R-spondin fusion colorectal cancer (CRC), whereas APC-mutant CRCs are resistant (Hinze et al., Cancer Discovery, 2020). To elucidate mechanisms of asparaginase resistance in APC-mutant CRC, we conducted an unbiased genome-wide loss-of-function CRISPR/Cas9 screen, which revealed that loss of G6PD confers marked asparaginase sensitivity. This finding was validated using independent CRISPR/Cas9 and shRNA approaches in multiple human CRC cell lines and mouse intestinal organoid models (P < 0.0001 for all comparisons). G6PD, the rate-limiting enzyme of the pentose phosphate pathway (PPP), was found to mediate asparaginase resistance through maintenance of NADPH production, whereas other PPP functions, such as ROS detoxification or ribonucleotide biosynthesis, were not implicated. Untargeted metabolomics indicated that G6PD loss impairs the urea cycle, and we found that G6PD inactivation selectively sensitized cells to ammonia toxicity but not to asparagine deprivation (P < 0.0001 for ammonia; P = n.s. for asparagine-free media). In G6PD-deficient cells, the toxicity of asparaginase was rescued by small molecules that enhance ammonia detoxification (P < 0.0001) or by knockdown of NMRAL1, a redox sensor that inhibits the urea cycle enzyme ASS under NADPH-limiting conditions (P < 0.05). To assess whether G6PD deficiency influences asparaginase toxicity in humans, we analyzed a cohort of 586 children with acute lymphoblastic leukemia (ALL) treated with asparaginase and with available assessments of erythrocyte G6PD activity. Seventeen patients had functional evidence of G6PD deficiency, and we observed no increase in grade ≥3 asparaginase-related toxicities among G6PD-deficient patients, either individually or in combination (P = n.s.). Finally, genetic inhibition of G6PD enhanced asparaginase efficacy in vivo in mouse CRC models (P < 0.01 and P < 0.05), and a tool G6PD inhibitor induced profound asparaginase sensitivity in patient-derived CRC organoids (P < 0.0001). Together, these findings reveal that G6PD deficiency unmasks a tumor-selective vulnerability to ammonia toxicity, which can be therapeutically exploited using asparaginase. This work provides a rationale for clinical testing of asparaginase in G6PD-deficient CRC and for the development of clinical G6PD inhibitors as combination partners for asparaginase therapy.