Background: Malnutrition is prevalent in GI malignancies and is associated with decreased survival, frequent hospitalizations and increased healthcare costs. Outside the perioperative setting, malnutrition rarely impacts treatment decisions. The subjective global assessment (SGA) is a validated tool that stratifies patients by nutritional status. We hypothesized that malnutrition independently predicts chemotherapy-related toxicity in patients with GI malignancies. Methods: All patients enrolled in the GI Cancer Registry which collects baseline demographic data, comorbid conditions and baseline nutrition status. In patients treated with chemotherapy, we reviewed dosing, schedule and toxicities over the first two months of treatment and compared results between well-nourished (SGA-A) and malnourished (SGA B/C) patients. Two-sample t-test or chi-square tests were used for comparisons as appropriate. Multivariable logistic regression analysis was used to estimate the independent effect of SGA status after controlling for demographic/prognostic factors. Results: We enrolled 184 patients between 5/2012 – 5/2013. The overall rate of malnutrition was 31.0%: 40.4% upper GI, 13.6% lower GI, 50.0% hepatobiliary. Malnourished patients received less chemotherapy (mean 63.9±23.8% vs. 81.4±18.8% of standard dose, p<0.0001) due to toxicity-related dose-reductions, even when stratified by tumor type or treatment type (based on toxicity risk). After controlling for age, gender, ECOG (≥1 vs. 0) and albumin in a multivariable logistic regression model, malnutrition was an independent predictor of receiving less than 80% of the standard dose (adjusted odds ratio = 2.30; 95% confidence interval = 1.07, 4.95, P=0.03). Malnourished patients had a slightly higher rate of hospitalization, not statistically significant (29.8% vs. 22.0%, p=0.26). Conclusions: In patients receiving chemotherapy for GI malignancies, malnutrition (defined by SGA B/C) is an independent predictor of toxicity-related chemotherapy dose-reductions. Prospective studies should evaluate whether nutrition interventions in malnourished patients decrease chemotherapy-related toxicities.