3 Background: Palliative chemotherapy is aimed at improving quality of life and increasing life expectancy, without curative intent. Toxicity during palliative treatment defeats the goal of care and increases healthcare cost. We describe the predictors of CRHs among cancer patients treated for palliative intent at a community cancer center. Methods: We conducted a nested case-control study of adult cancer patients who received chemotherapy from Jan 03 to Dec 11. Utilizing a pharmacy database we identified patients who had received chemotherapy for palliative intent. For quality measures cancer center prospectively collects data on all the patients who developed CRH, which was used to identify cases. Hospitalizations were classified as CRH, disease-related or unrelated by a multidisciplinary panel. We frequency matched 2 controls to per case on lines of their chemotherapy treatment (3 groups). We obtained odds ratios (OR) and 95% confidence intervals (CI) from a multivariable logistic regression model on this set of 199 cases and 398 controls to identify predictors of CRH. A two-sided p-value of 0.05 was used for all measures of statistical significance. Results: During the selected period 6,850 patients received chemotherapy, 2,559 (37.3%) for palliative intent. 230 (9%) of 2,559 developed CRH. 76.5 % of CRH happened during the first 3 cycles of chemotherapy, and the mean length of stay was 5 days. Significant predictors on multivariable regression were ECOG score (p = .03), Charlson score (p= .0018), cancer site (p <.006,) abnormal creatinine (p <.0001) and low albumin (p < .007). Conclusions: The results of this mature study demonstrate that patients treated with palliative chemotherapy have a substantial risk of severe hospital-requiring toxicity resulting in morbidity and cost. The finding that risk of severe toxicity is increased among patients with poor PS, multiple comorbidities and renal insufficiency suggests that some events may be avoidable. Furthermore, identified risk factors will enable the development and testing of a predictive model which could be used to identify patients at high risk of CRH prior treatment initiation.