Systematic literature reviews are common in social research for integrating and synthesising existing research. This paper argues that the outcomes of such reviews are affected by the choice of bibliographic databases. It presents evidence of substantial variation across three large electronic databases (Scopus, Web of Science and EBSCO) in a study on employee retention and staff turnover. It considers the specific articles, numbers returned, numbers shared across databases and perceived quality of journals hosting the retrieved articles. Results show that only 130 articles (5.7% of 2267 retrieved) were found common to all three databases, suggesting that decisions on how and where literature is retrieved can substantially affect the results of systematic reviews and meta-analyses. The findings caution against the use of single databases and claiming comprehensiveness. The paper reflects on how additional literature search methods (e.g., contacting experts, citation indices) and their sequence of use can affect systematic review quality.