Interpretations by physicians and those generated by electrocardiograph computer softwares have poor ability to recognize different types of supraventricular tachycardia (SVT). Therefore, we developed and tested a new SVT algorithm based on easily identifiable morphological characteristics and a simple dichotomous yes/no format regarding initial electrocardiographic manifestation and response pattern. The algorithm was then tested by medical house staff during the initial evaluation of 50 adult ED and cardiac intensive care unit patients suspected of having SVT. For a wide representation of SVTs, the new algorithm gave an overall diagnostic accuracy rate of 90%. Adenosine use was limited to 54% of the cases. No patient developed hemodynamic instability after algorithm-dictated interventions were carried out. Electrocardiograph computer-generated diagnoses correctly identified the specific type of SVT in 38% of the cases. This study shows the effectiveness of the proposed new algorithm in the rapid bedside evaluation and management of SVTs and confirms that computer-generated diagnoses are unreliable.