BACKGROUND: Cooling with water-flow blankets, which are difficult to manipulate and interfere with patients' care, may be ineffective in controlling fever. OBJECTIVE: To compare the effectiveness of cooling via convective airflow blankets with cooling via conductive water-flow blankets for treatment of fever in critically ill adults. METHOD: A 2-group experimental design was used to compare cooling via convection (n = 20) with cooling via conduction (n = 17) in critically ill adults with an infection-related fever of 38.5 degrees C or greater. Esophageal temperature was measured every 15 minutes until a temperature of 38.0 degrees C was reached or 8 hours had elapsed. Alternative cooling measures were withheld unless the temperature increased to more than 40.0 degrees C. Data on nurses' satisfaction were collected, and complications related to each cooling method were examined. RESULTS: Temperatures decreased more rapidly in the airflow group (mean decrease, 0.377 degree C/h) than in the water-flow group (mean decrease, 0.163 degree C/h). A temperature of 38.0 degrees C was achieved more often in the airflow group (75% vs 47.1%). Fever (temperature > 38.5 degrees C) recurred sooner in the water-flow group (6.6 hours) than in the airflow group (22.2 hours). Both methods were easy to use. Compared with the water-flow blanket, the airflow blanket was recommended for future use twice as often and interfered less with patients' care. CONCLUSIONS: In critically ill adults with an infection or a suspected infection, cooling with an airflow blanket is more effective and more preferred for cooling than is cooling with a water-flow blanket.