Background: It is well known that, at the time of the initial presentation of patients with non-small cell lung cancer, N-status is very important in accurate staging and avoiding inappropriate surgical procedures. Also, if mediastinal lymph node metastasis (N2) of non-small cell lung cancer (stage IIIA) is present, neoadjuvant (preoperative) chemora-diotherapy is indicated; therefore a quick and accurate N-status determination is critical before thoracotomy. We compared the usefulness and accuracy of touch imprint and frozen section analysis for this purpose.Methods: During 2006, fifty eight mediastinal lymph node specimens from 27 patients with lung cancer were excised. After touch imprint preparation, half of each lymph node was sent for frozen section and finally permanent histology.Results: Frozen section and touch imprint had only one and two false negative results, respectively, but neither had false positive reports. The sensitivity, specificity and accuracy of the frozen section evaluation was 95.5%, 100% and 98.2%, respectively, and those of touch imprint were, 90.9%, 100% and 96.4%. The differences in the sensitivity, specificity and accuracy between the two methods were not significant (p>0.05).Conclusion: Frozen section and touch imprint have similarly high accuracies for detecting mediastinal lymph node metastasis in lung cancer. However, since the touch imprint method is more rapid and simple, it maybe method of choice.