Background: Systematic review and meta-analysis of observational studies is getting more and more common as systematic review and meta-analysis of controlled clinical trials in order to answer the etiologic or effectiveness questions from the explosion information of scientific literatures. Cohort studies provide the second strongest evidence next to randomized controlled trials in evidence-based health care. In 1997, the workshop in Atlanta had proposed the concerns regarding the reporting of Meta-analysis Of Observational Studies in Epidemiology (MOOSE). However, no uniform checklist tool to assess the quality of cohort studies was recommended. Objective: The aim of this study is to evaluate the applicability of the Newcastle-Ottawa Scale (NOS) which is available in the supplementary material part of Cochrane Handbook for Systematic Review of Interventions. We analyze the items of NOS for rating quality of retrieved cohort studies evolving the predictors returning to work after traumatic limb injuries. Methods: A librarian conducted a database search of the PubMed/Medline electronic databases from inception to Feb 2010, using medical subheading term (MeSH) and free keyword text of the following search strategy: ((wounds and injuries) OR ( Wounds and Injuries [Mesh]) OR ( Cumulative Trauma Disorders [Mesh]) OR (cumulative trauma disorders) OR (injur* OR trauma OR fractur* OR limb injur* OR musculoskeletal injur*)) AND (return to work OR return-to-work OR work re-entry OR sickness leave). No language restriction was imposed. A hand search of the bibliographies of relevant papers was also carried out. Two reviewers independently performed initial screen and assessed of quality indicators after a consensus to the details of coding manual of NOS. Reliability of ratings of NOS items was calculated using kappas and reliability of the total (summed) score between two raters was calculated using intraclass correlation coefficients (ICC). Results: Our literature search uncovered 747 articles, irrelevant 479 and 139 of which were discarded after reviewing the titles and abstracts respectively, while 108 of which were abandoned after obtaining the full texts. One additional article was also identified for inclusion by hand searching the reference lists of obtained articles. Therefore, a total of 22 studies were candidates for the NOS appraisal by two independent raters. The kappa value for each of the 9 items (4 of selection, 2 of comparability, and 3 of outcome assessments) were ranged from 0.38 to 1.00. The ICC for the total score was 0.51 (95% confidence interval = 0.36 0.66) by Spearman s rho correlation. Conclusions: The reliability of ratings of NOS items varied from fair to almost perfect, and the correlation of the total NOS score between two rates was fair to good. However, there are still items (i.e., Demonstration that the outcome of interest was not present at start of study) might not be applicable for the appraisal of this research question since the outcome of returning to work will be always occurred after injuries. Acknowledgement: This study was supported by research grants from the Evidence-Based Medicine Committee of E-Da hospital.