BACKGROUND: Early red blood cell transfusions are a common treatment for adults hospitalized for sepsis without shock. However, their utility and association with mortality and costs have not been well studied. OBJECTIVES: To examine early transfusion rates for patients with sepsis treated outside intensive care units, and to find a correlation between transfusion rates and survival rates and costs. METHODS: Data were obtained from hospital members of the Premier Healthcare Alliance that admitted at least 50 adults with sepsis between January 1, 2006, and December 31, 2010. Early transfusion rates at each hospital were calculated as the observed incidence of allogeneic red blood cells administered by hospital day 2. A multivariable logistic regression model was constructed to estimate the expected or risk-adjusted transfusion rates, mortality rates, and costs. RESULTS: A total of 256 396 adults were hospitalized with sepsis without major bleeding or surgery at 364 US hospitals. Approximately 84% of all patients admitted with sepsis, without vasopressor therapy, were treated outside the intensive care unit (by day 2). The mean institutional early transfusion rate was 6.9%. After risk standardization, the median (interquartile range) transfusion rate was 6.7% (5.8%-7.6%), mortality rate was 15.5% (13.1%-18.1%), and costs were $13 333 ($11 939-$14 986). Early transfusion rates were not correlated with mortality but were modestly positively correlated with costs (Spearman ρ = 0.157; P = .003). CONCLUSIONS: Early transfusion rates during hospitalization for sepsis without shock varied widely across the hospitals. Transfusion rates were associated with increased costs but not with mortality rates.