Background: Patients with advanced GI cancer have a high prevalence of CACS. Medications used for CACS have limited efficacy, unwanted side effects, and high costs. We sought to describe the prescribing patterns of medications for CACS in patients with advanced GI cancer. Methods: We identified patients with advanced GI cancer (primary pancreatic, gastric, colorectal, and hepatobiliary) treated at the UM Fairview Medical Center, Minneapolis, MN from 2011-2019. We extracted sociodemographics, clinicopathological data (e.g. body mass index [BMI]), other treatments (e.g. nutrition referral), and prescriber specialty (e.g. oncology). The primary outcome was medication use for CACS: outpatient oral prescription of ≥7 days of megestrol, dronabinol, mirtazapine, olanzapine, or cyproheptadine as any formulation (e.g., tablet, solution) on a scheduled basis. Since medications can have overlapping indications, we clarified prescription indication through chart review if needed. We excluded dexamethasone given extremely non-specific use. Results: We identified 974 patients with advanced GI cancer (45.9 % women, 84.2% White, 4.7% Black, median age 65 years, 36.8% pancreatic cancer, median overall survival 212 days). Of these, 265 (27.2%) patients received at least 1 prescription for CACS (Table). There were 736 unique prescriptions (41.8% dronabinol, 26.9% mirtazapine, 22.4% olanzapine, 8.4% megestrol and 0.4% cyproheptadine). Brand-name formulations comprised 2.7% prescriptions. Most prescriptions were provided by oncologists (67.1%), followed by primary care practitioners (20.3%), and palliative care (6.0%). The median times from advanced cancer diagnosis to first prescription, and from last prescription to death were 73 and 50 days respectively. The median BMI (kg/m2) and serum albumin (g/dL) at time of first prescription were 24.3 (range 11.8-51.1) and 2.5 (range 0.7-4.5) respectively. Conclusions: Almost 1/3rd of patients with advanced GI cancers received medications for CACS. Prescribing was primarily driven by oncologists, and was common in the last weeks of life. Our findings can guide interventions to optimize CACS care by reducing reflexive medication prescribing for CACS, while underscoring the need to develop better pharmacologic and non-pharmacologic CACS treatments.