The implantable cardioverter defibrillator (ICD) offers life-saving therapies for primary and secondary prevention of sudden cardiac death in high-risk patients. However, ICD detection algorithms consistently misclassify a substantial proportion of supraventricular rhythms, thus carrying the risk for inappropriate therapies. Although single-chamber ICD (Sc-ICD) discrimination tools have been reported to provide high specificity in rejecting sinus tachycardia and atrial fibrillation with a relatively low ventricular rate, accurate recognition of atrial fibrillation with faster ventricular rates, atrial tachycardias, atrial flutter, and some reentrant tachycardias is still an issue. Dual-chamber ICDs (Dc-ICDs) are supposed to overcome specificity issues by enhancing detection algorithms with information derived from the atrial and ventricular timing relationship. The initial promise of Dc-ICDs was to improve detection specificity without compromising sensitivity, and to translate this advantage over Sc-ICDs in a more selective use of aggressive therapies. Despite this solid background, superiority of Dc- over Sc-ICDs has never been convincingly demonstrated. The present review focuses on the efficacy of contemporary ICD arrhythmia discrimination tools and appraises the so far reported evidence supporting the superiority of Dc-ICDs in preventing inappropriate therapies.