PURPOSE The purposes of this study were to evaluate the clinical course and to identify independent predictors of mortality in patients with cancer with sepsis. MATERIALS AND METHODS This is a secondary analysis of a prospective cohort study conducted at an oncological medical-surgical intensive care unit. Logistic regression was used to identify predictors of hospital mortality. RESULTS A total of 563 patients (77% solid tumor, 23% hematologic malignancies) were included over a 55-month period. The most frequent sites of infection were the lung, abdomen, and urinary tract; 91% patients had severe sepsis/septic shock. Gram-negative bacteria were responsible for more than half of the episodes of infection; 38% of patients had polymicrobial infections. Intensive care unit, hospital, and 6-month mortality rates were 51%, 65%, and 72%, respectively. In multivariate analyses, sepsis in the context of medical complications; active disease; compromised performance status; presence of 3 to 4 systemic inflammatory response syndrome criteria; and the presence of respiratory, renal, and cardiovascular failures were associated with increased mortality. Adjusting for other covariates, patients with non-urinary tract infections, mostly represented by patients with pneumonia and abdominal infections, had worse outcomes. CONCLUSIONS Sepsis remains a frequent complication in patients with cancer and associated with high mortality. Our results can be of help to assist intensivists in clinical decisions and to improve characterization and risk stratification in these patients.