PIK3CA encodes the p110α catalytic subunit of PI3K and is frequently mutated in human cancers, including ∼30% of colorectal cancer. Oncogenic mutations in PIK3CA render colorectal cancers more dependent on glutamine. Here we report that the glutaminase inhibitor CB-839 preferentially inhibits xenograft growth of PIK3CA-mutant, but not wild-type (WT), colorectal cancers. Moreover, the combination of CB-839 and 5-fluorouracil (5-FU) induces PIK3CA-mutant tumor regression in xenograft models. CB-839 treatment increased reactive oxygen species and caused nuclear translocation of Nrf2, which in turn upregulated mRNA expression of uridine phosphorylase 1 (UPP1). UPP1 facilitated the conversion of 5-FU to its active compound, thereby enhancing the inhibition of thymidylate synthase. Consistently, knockout of UPP1 abrogated the tumor inhibitory effect of combined CB-839 and 5-FU administration. A phase I clinical trial showed that the combination of CB-839 and capecitabine, a prodrug of 5-FU, was well tolerated at biologically-active doses. Although not designed to test efficacy, an exploratory analysis of the phase I data showed a trend that PIK3CA-mutant patients with colorectal cancer might derive greater benefit from this treatment strategy as compared with PIK3CA WT patients with colorectal cancer. These results effectively demonstrate that targeting glutamine metabolism may be an effective approach for treating patients with PIK3CA-mutant colorectal cancers and warrants further clinical evaluation. SIGNIFICANCE: Preclinical and clinical trial data suggest that the combination of CB-839 with capecitabine could serve as an effective treatment for PIK3CA-mutant colorectal cancers.