Objective: To estimate the frequency of therapeutic errors and to evaluate the diagnostic accuracy in the recognition of shockable rhythms by automated external defibrillators., Design: A retrospective descriptive study., Setting: Nine basic life support units from Biscay (Spain)., Participants: Included 201 patients with cardiac arrest, since 2006 to 2011., Measurements: The study was made of the suitability of treatment (shock or not) after each analysis and medical errors identified. The sensitivity, specificity and predictive values with 95% confidence intervals were then calculated., Results: A total of 811 electrocardiographic rhythm analyses were obtained, of which 120 (14.1%), from 30 patients, corresponded to shockable rhythms. Sensitivity and specificity for appropriate automated external defibrillators management of a shockable rhythm were 85% (95% CI, 77.5% to 90.3%) and 100% (95% CI, 99.4% to 100%), respectively. Positive and negative predictive values were 100% (95% CI, 96.4% to 100%) and 97.5% (95% CI, 96% to 98.4%), respectively. There were 18 (2.2%; 95% CI, 1.3% to 3.5%) errors associated with defibrillator management, all relating to cases of shockable rhythms that were not shocked. One error was operator dependent, 6 were defibrillator dependent (caused by interaction of pacemakers), and 11 were unclassified., Conclusions: Automated external defibrillators have a very high specificity and moderately high sensitivity. There are few operator dependent errors. Implanted pacemakers interfere with defibrillator analyses., (Copyright © 2012 Elsevier España, S.L. All rights reserved.)