Introduction: Ablation of muscular fascicles around the ostium of pulmonary veins (PVs) resulting in electrical isolation of the veins may prove to be an effective treatment for atrial fibrillation (AF). Correctly discriminating atrial and PV potentials is necessary to effectively isolate PVs from the left atrium in patients with paroxysmal AF. Methods and Results: A training set of 151 electrode recordings obtained from 10 patients with AF was used to develop an algorithm to discriminate atrial and PV potentials. Bipolar electrograms were collected from a multielectrode basket catheter placed sequentially into each PV. Amplitude, slope, and normalized slopes of both bipolar and quadripolar electrograms (difference between adjacent bipoles) were entered into a binary logistic regression model. A receiver operating characteristic curve was used to define a threshold able to effectively discriminate atrial and PV potentials. The normalized slopes of both domains, bipolar and quadripolar, produced a logistic function that discriminated atrial and PV potentials against a threshold (0.38) with 97.8% sensitivity and 94.9% specificity. The algorithm then was evaluated on a test set of 214 electrode recordings from four patients who also had paroxysmal AE These patient electrograms also were evaluated by two independent electrophysiologists. The algorithm and electrophysioiogists matched identification of activation origin in 84% of electrograms. Conclusion: Atrial and PV potentials acquired from a multielectrode basket catheter can be discriminated using the normalized slopes of bipolar and quadripolar electrograms. These additional parameters need to be included by physicians determining the preferential ablation site within PVs.