Résumé : Aim: Spontaneous alterations in temperature homeostasis after cardiac arrest (CA) are associated with worse outcome. However, it remains unclear the prognostic role of temperature variability (TV) during cooling procedures. We hypothesized that low TV during targeted temperature management (TTM) would be associated with a favourable neurological outcome after CA. Methods: We reviewed data from all comatose patients after in-hospital or out-of-hospital CA admitted to our Department of Intensive Care between December 2006 and January 2014 who underwent TTM (32-34. °C) and survived at least 24. h. We collected demographic data, CA characteristics, intensive care unit (ICU) survival and neurological outcome at three months (favourable neurological outcome was defined as cerebral performance category 1-2). TV was expressed using the standard deviation (SD) of all temperature measurements during hypothermia; high TV was defined as an SD >1. °C. Results: Of the 301 patients admitted over the study period, 72 patients were excluded and a total of 229 patients were studied; 88 had a favourable neurological outcome. The median temperature on ICU admission was 35.8 [34.9-36.9]. °C and the median time to hypothermia (body temperature <34. °C), was 4 [3-7] h. Median TV was 0.9 [0.6-1.0]. °C and 57 patients (25%) had high TV. In multivariable logistic regression, witnessed CA, ventricular fibrillation/tachycardia and previous neurological disease were independent risk factors for high TV. Younger age, bystander cardiopulmonary resuscitation, shorter time to return of spontaneous circulation, cardiac origin of arrest, shockable rhythm and longer time to target temperature were independent predictors of favourable neurological outcome, but TV was not. Conclusions: Among comatose survivors treated with TTM after CA, 25% of patients had high TV; however, this was not associated with a worse neurologic outcome.