Objective: We evaluated our results from our prospective database to identify possible modifications that may improve our fast-tracking protocols in selected high-risk patients. Methods: We conducted a retrospective study of a prospective database. Using multivariable regression, we identified several patient characteristic that predicted failure to fast-track owing to increased morbidity. We modified our fast-tracking algorithm by substituting pain pumps for epidurals in elderly patients (>70 years). In addition, patients with a body mass index greater than 35 had increased aspiration precautions. Patients with poor pulmonary function (ratio of forced expiratory volume in 1 second to forced vital capacity and/or diffusing capacity/alveolar volume < 45%) underwent increased respiratory treatments and more aggressive ambulation. Differences in outcomes between groups were compared after adjusting for differing baseline patient characteristics, including use of a propensity score. Results: A total of 2895 patients underwent elective pulmonary resection before the algorithm modifications (January 1997-December 2001) and 3252 patients afterward (January 2002-July 2007) by one surgeon. The length of stay was reduced by the protocol changes from 6.7 to 4.9 days (P = .024) in elderly patients, from 5.7 to 4.8 days in obese patients, and from 6.2 to 4.3 days (P = .008) in those with poor pulmonary function. Morbidity was reduced from 26% to 17% in elderly patients (P = .046), from 29% to 20% (P = .027) in obese patients, and from 45% to 23% in those with poor pulmonary function. Overall mortality was also reduced 4.0% to 2.1% (P = .014). Conclusion: A prospective database provides important information that can lead to improvement in patient care by identifying specific complications. High-risk patients such as the elderly, the obese, and those with poor pulmonary function can safely undergo pulmonary resection and have a shorter hospital stay. © 2009 The American Association for Thoracic Surgery.