This chapter evaluates the diagnostic accuracy of positron emission tomography in non-small cell lung cancer (NSCLC). In NSCLC, evaluation of the extent of disease in the mediastinum is an important step in deciding about further therapy. In general, patients with preoperative identifiable N2 disease are excluded from thoracotomy. Conventional work-up of the mediastinum often includes noninvasive imaging with computed tomography (CT), followed by invasive exploration through mediastinoscopy. In a meta-analysis of 18 studies, including more than a thousand patients, the negative predictive value of PET was found to be 93%, which is at least as good as that of mediastinoscopy in daily clinical practice. In oncology, stored surgical specimens can also provide useful information on the potential impact of a diagnostic test. Decision analysis models the cost-effectiveness of a new diagnostic device. The model can combine results of clinical studies that cover different health-care steps. In the presence of many alternative diagnostic strategies, decision analysis can help to identify the most promising diagnostic tests or algorithms for further research. Several aspects are of particular importance for randomized controlled trials (RCT) involving diagnostic imaging. In therapeutic RCT the outcome is usually measured in terms of patient mortality and morbidity. However, diagnostic tests serve to allocate appropriate therapy to patients. A reasonable outcome measure for such studies is the extent to which appropriate therapy is applied, depending on the condition that the new test does not alter the definitions of staging and treatment for each stage that is already established.