Charles et al have nicely shown that implementation of clinical guidelines derived from evidence-based medicine can change physician behavior. Indeed, evidence-based medicine is slowly being adopted in areas where such data exist. The crystalloid-colloid argument for resuscitation of trauma and burn patients has long been over, with the exception of instances in which lower volumes would be logistically helpful on the battlefield. However, the use of colloids has its proponents, now mostly anesthesiologists, who use it in the intraoperative and perioperative periods. Hetastarch rather than albumin is used because of the expense of the latter. The idea that colloid administration would result in less tissue edema has not been borne out experimentally because third spacing of colloids eventually occurs. The Cochrane database has repeatedly shown no difference in mortality with albumin use, and more recently the SAFE trial has also shown no benefit from albumin in the critical care patient. Despite these data, albumin use has been excessive in some centers. Following an educational program that included discussion during rounds, journal clubs, and didactic sessions, albumin use was reduced without any change in ICU mortality during the study period of 1 year. Associated costs were reduced as well, and illness severity, as measured by APACHE III scores, was similar during the observation period. This study is an excellent example of how such evidence-based medicine can be implemented locally, but it takes a champion of the cause and a team concept as well for institutional change. Although the study was reliant on historical controls and not randomized, it was prospective and demonstrated a marked reduction in albumin use that persisted during the following year. One wonders if there will be much use for albumin in this setting now that the data are compelling against its use in the ICU.