Background: Type 2 Diabetes (T2D) is prevalent among patients with cancers of the stomach, liver, pancreas and biliary tree. Improved control of T2D preoperatively and perioperatively may mitigate known risks for infection, wound healing and length of stay. We estimate the association of T2D control on surgical outcomes and identify groups that may benefit from improved T2D management. Methods: Utilizing SEER linked with Medicare administrative claims, we identified patients diagnosed with gastric, hepatic, pancreatic and biliary cancers (2008-2015) who underwent open operations with curative intent ≤6mo of diagnosis. Based on utilization 1y prior to surgery, T2D history was characterized as non-diabetic, controlled (T2D with no complication claims), history of uncontrolled T2D claims, and T2D with hospitalization or ED visit for T2D. Multivariate regression analyses examined risk factors for a history of uncontrolled/hospitalizations/ED for T2D, including age at cancer diagnosis, sex, race/ethnicity, census tract poverty, rural status, and Elixhauser Comorbidity (excluding T2D). Multivariate models examined T2D status (ref: non-T2D) and risk of each surgical outcomes (length of stay, 30d readmission, complications, 30d ED visits) controlling for similar factors including cancer and stage. Results: Among 8,015 patients, 52% were male, 68% were non-Hispanic white and the median age at cancer diagnosis was 76y. The most frequent cancer diagnosis was gastric (41%), followed by pancreatic (38%), biliary (13%), and hepatic (8%). The majority of patients were diagnosed with Stage II (44%) disease. Almost half of patients had a history of T2D prior to cancer surgery (47.5%). Among T2D, 60.6% were controlled, 34% uncontrolled claims, and 5.4% with history of T2D hospitalization/ED. Risk factors for any uncontrolled status included non-Hispanic Black and Hispanic race/ethnicity (RRs 1.21, 1.31, all p<0.01) and having 2+ additional comorbidities (RR 1.8, p<0.001). All T2D groups had an increased risk of 30d readmission compared to non-T2D in multivariate analysis; particularly those with a history of T2D hospitalization/ED (RR 1.47, p<0.001) and uncontrolled claims (RR 1.17, p=0.01). Prior T2D hospitalization/ED (RR 1.49, p<0.001) and uncontrolled claims (RR 1.18, p<0.01) increased the risk of 30d ED visits compared to non-T2D. No differences between controlled and non-T2D for readmission or ED visits were observed. For readmission and ED visit, a history of T2D hospitalization/ED had significantly greater risk compared to controlled T2D. Conclusions: Poor T2D control prior to surgery for UGI cancers is associated with an increased risk of readmissions and ED visits, while those with controlled T2D did not experience the excess risk. Patients of minority race/ethnicity were at the greatest risk of both uncontrolled status and poor outcomes. Evidence supports increasing diabetes management prior to surgery.