Surgical atrial fibrillation ablation (SAFA) has not achieved the efficacy of Cox's original maze procedure, although technical improvements continue to be made. It is possible that biologic factors determine SAFA success. Therefore we examined how patient-specific characteristics affected SAFA success in 353 atrial fibrillation (AF) patients who underwent SAFA at a single institution. Among these, 257 (72.8%) had continuous AF and 96 (27.2%) had intermittent AF. For 297 patients (84.1%) postoperative follow-up was >3 months. We compared SAFA success in patients whose procedure involved only pulmonary vein isolation with those whose procedure involved extensive lesion sets. Multivariate analysis included AF duration, left atrial size, preoperative atrial flutter, concomitant procedures, lesion sets, and energy source. Early SAFA success was classified as freedom from AF between postoperative months 3 and 6, and intermediate success between postoperative months 6 and 12. Receiver-operating characteristic (ROC) curves and stratum-specific likelihood ratios (SSLR) were generated to compare intermediate failure by left atrial size (LAS) thresholds. SAFA was more successful in the intermittent than the continuous AF group (n = 66, 86% vs n = 165, 71%; P = .014). When pulmonary vein isolation alone was compared to more extensive lesion sets, there was no difference in success in the intermittent (34, 91% vs 32, 81%; P = .24) or continuous groups (67, 73% vs. 98, 69%; P = .603). Success for intermittent AF patients was not correlated with variables considered; in continuous AF patients, predictors included presence of concomitant mitral valve repair/replacement (P = .075), decreasing LAS (P = .025) and absence of preoperative atrial flutter (P = .001). In the continuous AF group, ROC curves and corresponding areas under the curve (AUC) were 0.60 (0.50-0.71) for failure at 6 months to 1 year. SSLR analysis generated 2 strata for LAS: <8 cm with SSLR = 0.87 (0.74-1.0) and ≥8 cm SSLR = 2.98 (1.07-8.3). In patients with intermittent AF, SAFA achieved acceptable results regardless of tested preoperative and intraoperative variables. In continuous AF, patient-specific characteristics affected success more than intraoperative variables. Failure was more than 3-fold greater in continuous AF patients with an LAS ≥8 cm. In both patient types, more extensive lesion sets were not shown to improve outcomes. Future improvements in SAFA may depend on pharmacologic and/or surgical substrate modification. © 2007 Forum Multimedia Publishing, LLC.