Properly conducted meta-analyses that are based on systematic reviews of the literature allow the conclusive synthesis of accumulating scientific evidence. Systematic reviews, with or without meta-analyses, offer a more objective appraisal of the available evidence compared with traditional narrative reviews. Combining data from independent studies using meta-analytic methods can improve statistical precision, but cannot prevent bias as such. The validity of meta-analyses depends on the methodological quality of the included studies, the eligibility criteria used for the meta-analysis, and the various reporting biases. In this Review we examine the analytical strengths of, and the main problems encountered by, both systematic reviews and meta-analyses, focusing on how to best assess the validity of each for the practicing clinician.