In the differential diagnosis of broad-complex tachycardia, the most important decision is whether or not the tachycardia is ventricular, since this type carries the worst prognosis. However, the rules for a diagnosis of ventricular tachycardia are so complex that they are not satisfied in many cases, and the default diagnosis, supraventricular tachycardia, is erroneously accepted. We sought to reverse this strategy; unless simple rules for a positive diagnosis of supraventricular tachycardia were satisfied, ventricular tachycardia was diagnosed by default. The criterion for a diagnosis of supraventricular tachycardia was electrocardiographic (ECG) findings typical of bundle branch block (left = rS or QS wave in leads V1 and V2, delay to S wave nadir < 70 ms, and R wave and no Q wave in lead V6; right = rSR' wave in lead V1 and an RS wave in lead V6, with R wave height greater than S wave depth). Twelve-lead ECGs were done for 102 consecutive patients with broad-complex tachycardia (QRS width > 110 ms). Two observers, who were unaware of definitive diagnoses validated by electrophysiology, by our diagnostic rules made correct diagnosis of ventricular tachycardia in 63 and 62 of 69 patients, respectively, and correct diagnoses of supraventricular tachycardia in 28 and 22 of 33 patients (sensitivity for ventricular tachycardia 90% and 91%, specificity 67% and 85%). One observer then sought independent P waves in cases diagnosed as supraventricular tachycardia; sensitivity for the diagnosis of ventricular tachycardia rose to 96%, with a specificity of 64%. These criteria, which require only knowledge of typical bundle branch block patterns, were highly sensitive for the important diagnosis of ventricular tachycardia.