Prediction of accessory pathway location before radio-frequency ablation has become increasingly important for patients with Wolf-Paridnson-White syndrome. However, existing electrocardiographic (ECG) criteria for localization of accessory pathways have several limitations, and the polarity of delta waves has not been well defined. In the present study, 369 patients with a single anterogradety conducting accessory pathway who underwent successful radiofrequency ablation were included. The polarity of delta waves was defined and categorized in detail, and various ECG characteristics of the most preexcited QRS complexes were examined and compared with QRS complexes after successful ablation in the initial 182 patients, which included morphology and polarity of delta waves, initial 20, 40, and 60ms segments of the preexcited QRS complex, R/S ratio in the precordial leads, R/S ratio in the frontal leads, delta wave axis in the frontal plane, polarity of delta waves in the frontal leads, and polarity of delta waves in the precordial leads. The polarity of the initial 40ms segment of the most preexcited QRS complexes in each of the frontal leads, and the polarity of the initial 60ms segment of the most preexcited QRS complex in each of the precordial leads proved to be the best representatives of delta wave polarity in the respective leads. The most discriminative characteristics were combined to form the following algorithm: step 1, analysis of R/S ratio in V₂; step 2, existence of positive delta wave in lead III (initial 40ms); step 3, existence of positive or negative delta waves in V₁, (initial 60ms); and step 4, delta wave polarity in aVF or analysis of R/S ratio in V₁. The new algorithm is featured by its easiness to be applied because only 4 ECG leads and 4 steps are required, and the same ECG characteristic is used in the same step in both arms of the algorithm. Together with 3 other criteria (those of Arruda, Filzpatrick, and Xie), it was prospectively evaluated in 187 subsequent patients, and it turned out to be more accurate (93%) than other algorithms (86%, 86%, 85%, respectively; all p values <0.05). We conclude that this new algorithm, through a comprehensive analysis of delta waves and an R/S ratio in a large group of patients, offers an accurate criterion for predicting accessory pathway location.