There is controversy over whether to scan extrathoracic sites for metastases in patients with non-small cell lung cancer (NSCLC). We tested the efficiency of clinical factors to determine whether metastasis has occurred, and whether routine scanning for NSCLC is required. Nine hundred and forty five patients scanned for extrathoracic metasates were included. Clinical factors indicating metastasis were determined using multivariate analysis. Of the 945 cases, 377 (39.9%) had metastasis. Bone metastases were determined by focal skeleton pains, elevated serum alkaline phosphatase levels, adenocarcinoma, KPS</=70, sensitivity of 90.6, specificity of 12.7, PPV of 16.3, NPV of 87.8, and silent metastases rate (SMR) of 9.4%. Brain metastases were determined by neurological symptoms, adenocarcinoma, hematocrite <40 for men and <35 for women, KPS</=70, sensitivity of 89.9, specificity of 7.9, PPV of 9.2, NPV of 88.3, and SMR of 10.1%. Abdominal metastases were determined by abdominal pain/tension, hepatomegaly, elevated GGT levels, serum LDH levels >500 IU, a N2 or N3 case, KPS</=70, sensitivity of 95.9, specificity of 7.1, PPV of 13.3, NPV of 92.1 and SMR of 4.1%. Of the 224 patients with stage I and II disease, 73 had metastasis with a rate of 10.9% silent metastasis. We concluded that routine scanning of NSCLC for staging is necessary.