Background: Collections of meta-analyses assembled in meta-epidemiological studies are used to study associations of trial characteristics with intervention effect estimates. However, findings are not consistent across studies. A recent study that combined data from six meta-epidemiological studies identified overlap between the contributed systematic reviews, meta-analyses and randomised controlled trials (RCTs). Objectives: To combine data from ten meta-epidemiological studies into a single database, and to derive a dataset without overlap between included meta-analyses and RCTs. Methods: Data, supplied in various forms, were combined in one Access database. The design allowed for RCTs to be contained in different meta-analyses, multiple meta-analyses in systematic reviews, overlapping meta-analyses between systematic reviews and multiple references to the same RCT or review. Where available, unique identifiers (PMID, EMBASE ID, or ISI ID) were assigned to each reference, which was then linked to RCTs and systematic reviews. Unindexed and unpublished RCTs were assigned an alternative identifier. Results: The final database structure has six tables containing information on references, RCTs, trial results, meta-analyses, systematic reviews and relationships between the tables. The final table contains identifiers necessary for linking between tables. The combined dataset initially contained 455 meta-analyses from 428 systematic reviews, contributing 4874 trial results. Of these, 262 meta-analyses were unique - the remaining 193 shared at least one trial with another meta-analysis. Based on inspection of sets of overlapping meta-analyses, 73 systematic reviews and 89 meta-analyses containing 1344 trial results were removed. An additional 20 duplicated trial results were removed from 14 further metaanalyses where overlap was small. The final database contains 355 systematic reviews, 366 meta-analyses and 3510 unique trial results. Conclusions: The data structure allowed construction of a unique database that will be used to examine combined evidence on sources of bias in RCTs. The lessons learned during this process may be of use to The Cochrane Collaboration, since currently there is no way to examine overlap between different reviews in the Cochrane Database of Systematic Reviews.