Background: Dihyropyrimidine dehydrogenase (DPD) deficiency is present in 3-5% of patients, and is associated with substantially increased risk of severe and/or fatal toxicity during standard-dose FP chemotherapy. Genotyping of pathogenic DPYD variants is a readily available screening test for DPD deficiency, and prospective studies show that dose-reduced FP chemotherapy can be used safely in heterozygous DPYD variant carriers. Methods: Following a sentinel toxicity event the GI medical oncology group at the Norris Cotton Cancer Center adopted a shared practice of routine screening for pathogenic DPYD gene variants prior to FP chemotherapy (5-FU or capecitabine). Screening procedures involved physicians, NP/PAs, nurses, pharmacists, and schedulers. Testing was completed at a send-out lab until late 2020, when an in-house test became available. The current test panel evaluates for 3 gene variants: c.1905+1G > A (*2A), c.1679T > G (*13), and c.2846A > T. We report on the sustainability and clinical outcomes of DPYD gene variant screening. We identified all patients starting new FP-containing intravenous chemotherapy regimens (e.g., FOLFOX, CAPOX) for treatment of GI cancer at two sites (LEB & STJ) between Jan. 2019 and May 2021. We used electronic medical records to evaluate for completion of DPYD genotyping, and we describe the prevalence and management of DPYD gene variant carriers. Results: We identified 333 patients starting FP-containing chemotherapy regimens during the study period, including 287 patients without prior history of FP chemotherapy. Screening with DPYD genotyping was completed in 228 of 287 eligible patients (79%). Screening rates increased from 34% in Q1 of 2019 to 90% in Jan-May 2021. Five GI oncology sub-specialists accounted for 89% of screen-eligible patients and 96% of completed tests, but 10 unique physicians ordered ≥1 test. Of 228 screened patients, six (2.6%) were heterozygous carriers of pathogenic DPYD gene variants (*2A [2 patients], *13 [1], and c.2846A > T [3]). Variant carriers started FP chemotherapy with a 33-50% reduction. Two of six patients required further dose reduction due to FP-related toxicity (grade 4 neutropenia, grade 3 diarrhea). All evaluable variant carriers completed planned initial treatment. Implementation challenges included variable insurance coverage of DPYD genotyping, site-specific test ordering and reporting processes, and inconsistent turn-around time for send-out testing (resolved with on-site testing). Conclusions: Routine screening for pathogenic DPYD gene variants prior to FP chemotherapy is feasible and sustainable in the U.S. DPYD genotyping coupled with chemotherapy dose reductions for DPYD variant carriers facilitates safe and timely completion of planned chemotherapy treatments.