Recent studies suggest using balanced crystalloids (lactated Ringer's solution or Plasma-Lyte A) rather than saline (0.9% sodium chloride) may improve outcomes for patients with sepsis in the emergency department (ED) and intensive care unit (ICU).What is the relative impact on sepsis outcomes of fluid composition during early resuscitation in the ED versus after ICU admission?and Methods: We performed a secondary analysis of the Isotonic Solutions and Major Adverse Renal Events Trial (SMART) dataset examining medical ICU patients with a diagnosis of sepsis (n=1,641). SMART was a cluster-crossover trial comparing balanced crystalloids versus saline among critically ill adults. During the first seven months of SMART, fluid choice was controlled only in the ICU ("ICU only period"). In the final 15 months, fluid choice was coordinated between the ED and ICU ("ED & ICU period"). We performed logistic regression modeling for 30-day in-hospital mortality with an interaction term between randomized group (balanced crystalloids vs saline) and study period ("ICU only period" vs "ED & ICU period").A total of 367 patients with sepsis were enrolled during the ICU only period and 1,274 during the ED & ICU period. Thirty-day in-hospital mortality occurred in 47 of 142 patients (33.1%) in the balanced crystalloid group versus 74 of 225 patients (32.9%) in the saline group during the ICU only period (odds ratio, 1.14; 95% CI, 0.70-1.88) and 170 of 682 patients (24.9%) in the balanced crystalloid group versus 181 of 592 patients (30.6%) in the saline group in the ED & ICU period (odds ratio, 0.68; 95% CI, 0.52-0.89) (P value for interaction = 0.07) - consistent with a beneficial effect of balanced crystalloid primarily in the ED & ICU period.Among patients with sepsis, the effect of balanced crystalloids versus saline on mortality was greater for patients for whom fluid choice was controlled starting in the ED compared to starting in the ICU.