Introduction The goal of a systematic review is to compile and summarise all relevant reports on one well-defined topic. Selection bias is minimised by carefully defining search strategies, keywords, and eligibility criteria. However, remaining sources of bias may lead to the omission or under representation of relevant literature.

Methods The objective of this study was to identify and classify sources of selection bias in systematic reviews. The search methods and stated limitations of more than 50 systematic reviews were examined and catalogued.

Results Some forms of bias result from systemic barriers in the research dissemination process. These include (1) pipeline biases, such as those that hinder the dissemination of null association reports, and (2) indexing biases, such as the exclusion of non-English journals from major abstract databases. Other forms of bias result from methodological choices made by researchers. These include (3) search biases, such as those related to the use of hand-searches and grey literature searches, and (4) eligibility biases, such as those based on the language of publication or on particular indicators of quality, which may result in the exclusion of non-English reports and small-scale studies from regions with less developed health research infrastructure.

Conclusion A systematic review must set strict inclusion criteria, but the process used to select eligible studies may introduce a wide variety of biases. Awareness of the factors that restrict the comprehensiveness of systematic reviews will allow researchers to weigh and address these limitations. Improved systematic reviews will form a stronger foundation for evidence-based policy.