The use of CEA as a prognostic and predictive marker in patients with lung cancer is widely debated. The aim of this review was to evaluate the results from studies made on this subject. Using the search words "CEA", "tumor markers in lung cancer", "prognostic significance", "diagnostic significance" and "predictive significance", a search was carried out on PubMed. Exclusion criteria was articles never published in English, articles before 1981 and articles evaluating tumor markers in lung cancer not involving CEA. Initially 217 articles were found, and 34 were left after selecting those relevant for the present study. Four of these included both Non-Small Cell Lung Cancer (NSCLC) and Small Cell Lung Cancer (SCLC) patients, and 31 dealt solely with NSCLC patients. Regarding SCLC no studies showed that serum level of CEA was a prognostic marker for overall survival (OS). The use of CEA serum level as a prognostic marker in NSCLC was investigated in 23 studies and the use of CEA plasma level in two. In 18 (17 serum, 1 plasma) of these studies CEA was found to be a useful prognostic marker for either OS, recurrence after surgery or/and progression free survival (PFS) in NSCLC patients. Interestingly, an overweight of low stage (stage I-II) disease and adenocarcinoma (AC) patients were observed in this group. The remaining 7 studies (6 serum, 1 plasma) contained an overweight of patients with squamous carcinoma (SQ). One study found evidence for that a tumor marker index (TMI), based on preoperative CEA and CYFRA21-1 serum levels, is useful as a prognostic marker for OS in NSCLC. Six studies evaluated the use of CEA as a predictive marker for risk of recurrence and risk of death in NSCLC patients. Four of these studies found, that CEA was useful as a predictive marker for risk of recurrence and risk of death measured over time. No studies found CEA levels useful as a diagnostic marker for lung cancer. With regard to NSCLC the level of CEA measured in tumor tissue in NSCLC patients, were not of prognostic, diagnostic or predictive significance for OS or recurrence after treatment. In one study CEA level was measured in Pleural Lavage Fluid (PLF) it was here found to be useful as prognostic markers for overall survival (OS) after surgery. In conclusion serum level of CEA carries prognostic and predictive information of risk of recurrence and of death in NSCLC independent of treatment or study design. The observation that TMI index could be a potential prognostic marker for OS in NSCLC is interesting. Future studies may benefit from evaluating more than one marker at a time, which may possibly create a more precise index for prognosis and recurrence in lung cancer, than is possible by the use of single biomarkers. Copyright © 2011 Elsevier Ireland Ltd. All rights reserved.