To determine whether plasma cholesterol concentrations in dogs with sepsis is associated with morbidity or in-hospital mortality. Retrospective cohort study from 2005-2015. Two private referral centers. Fifty-one dogs diagnosed with sepsis. None. Dogs were classified as septic if they displayed ≥2 criteria of the systemic inflammatory response syndrome in conjunction with a documented underlying infectious cause. Dogs were excluded if they had been diagnosed previously with any concurrent illness reported to alter plasma cholesterol concentrations. Plasma cholesterol concentrations at the time of sepsis diagnosis were statistically analyzed for association with morbidity, as measured by the presence of organ dysfunction, the number of dysfunctional organs, duration of hospitalization, cost of hospitalization, and in-hospital mortality. Twenty-eight (55%) dogs survived to discharge, 15 (29%) were euthanized during hospitalization, and 8 (16%) died despite treatment. While median cholesterol concentrations were significantly different when comparing survivors to discharge versus nonsurvivors who died naturally despite treatment (P = 0.0245), they were not significantly different when comparing survivors to all nonsurvivors (P = 0.1821). Receiver operating characteristic curve analysis showed a cholesterol cutoff of 4.5 mmol/L (174 mg/dL) with a sensitivity of 75% and a specificity of 50% for predicting in-hospital mortality. For surviving dogs, plasma cholesterol concentrations were not associated with increased length of hospital stay. Number of dysfunctional organs and plasma cholesterol concentration were the 2 most significant individual predictors for survival, and when incorporated into a multivariate logistic regression model used for prediction, the model yielded a sensitivity of 94% and specificity of 63%. Plasma cholesterol concentration can provide prognostic information in dogs with sepsis. Further prospective studies investigating the role of cholesterol in sepsis are needed.