Objective: Accurate preoperative staging is essential for the optimal management of patients with lung cancer. An important goal of preoperative staging is to identify mediastinal lymph node metastasis. Computed tomography and positron emission tomography may identify mediastinal lymph node metastasis with sufficient sensitivity to allow omission of mediastinoscopy. This study utilizes our experience with patients with clinical stage I lung cancer to perform a decision analysis addressing whether mediastinoscopy should be performed in clinical stage I lung cancer patients staged by computed tomography and positron emission tomography. Methods: We retrospectively reviewed our thoracic surgery database for cases between May 1999 and May 2004. Patients deemed clinical stage I by computed tomography and positron emission tomography were chosen for further study. Individual computed tomography, positron emission tomography, and operative and pathology reports were reviewed. The postresection pathologic staging and long-term survival were recorded. A decision model was created using TreeAgePro software and our observed data for the prevalence of mediastinal lymph node metastases and for the rate of benign nodules. Data reported in the literature were also utilized to complete the decision analysis model. A sensitivity analysis of key variables was performed. Results: A total of 248 patients with clinical stage I lung tumors were identified. One hundred seventy-eight patients (72%) underwent mediastinoscopy before resection, and 5/178 (3%) showed N2 disease. An additional 9 patients were found to have N2 metastasis in the final resected specimen, resulting in a total of 14/248 patients (5.6%) with occult mediastinal lymph node metastases. Benign nodules were found in 19/248 (8%) of patients. Decision analysis determined that mediastinoscopy added 0.008 years of life expectancy at a cost of $250,989 per life-year gained. The outcome was sensitive to the prevalence of N2 disease in the population and the benefit of induction versus adjuvant therapy for N2 lung cancer. If the prevalence of N2 disease exceeds 10%, the sensitivity analysis predicts that mediastinoscopy would lengthen life at a cost of less than $100,000 per life-year gained. Conclusion: Patients with clinical stage I lung cancer staged by computed tomography and positron emission tomography benefit little from mediastinoscopy. The survival advantage it confers is very small and is dependent on the prevalence of N2 metastasis and the unproven superiority of induction therapy over adjuvant therapy.