INTRODUCTION: Cardiopulmonary resuscitation (CPR) is often started irrespective of comorbidity or cause of arrest. We aimed to determine the prevalence of perception of inappropriate CPR of the last cardiac arrest encountered by clinicians working in emergency departments and out-of-hospital, factors associated with perception, and its relation to patient outcome. METHODS: A cross-sectional survey was conducted in 288 centres in 24 countries. Factors associated with perception of CPR and outcome were analyzed by Cochran-Mantel-Haenszel tests and conditional logistic models. RESULTS: Of the 4018 participating clinicians, 3150 (78.4%) perceived their last CPR attempt as appropriate, 548 (13.6%) were uncertain about its appropriateness and 320 (8.0%) perceived inappropriateness; survival to hospital discharge was 370/2412 (15.3%), 8/481 (1.7%) and 8/294 (2.7%) respectively. After adjusting for country, team and clinician's characteristics, the prevalence of perception of inappropriate CPR was higher for a non-shockable initial rhythm (OR 3.76 [2.13-6.64]; P < .0001), a non-witnessed arrest (2.68 [1.89-3.79]; P < .0001), in older patients (2.94 [2.18-3.96]; P < .0001, for patients >79 years) and in case of a "poor" first physical impression of the patient (3.45 [2.36-5.05]; P < .0001). In accordance, non-shockable and non-witnessed arrests were both associated with lower survival to hospital discharge (0.33 [0.26-0.41]; P < 0.0001 and 0.25 [0.15-0.41]; P < 0.0001, respectively), as were older patient age (0.25 [0.14-0.44]; P < 0.0001 for patients >79 years) and a "poor" first physical impression (0.26 [0.19-0.35]; P < 0.0001). CONCLUSIONS: The perception of inappropriate CPR increased when objective indicators of poor prognosis were present and was associated with a low survival to hospital discharge. Factoring clinical judgment into the decision to (not) attempt CPR may reduce harm inflicted by excessive resuscitation attempts.