To illustrate the precision and sensitivity of identifying case reports of adverse events (AE) in MEDLINE and EMBASE, eight searching methods were created and applied to two systematic reviews of case reports. Search methods included use of indexing and text words for synonyms of AE and related terms, as well as drug, disease and study design.  All eight methods contained the drug and combinations of other search strings and ranged from broad to very specific. Searches were developed for MEDLINE and EMBASE; duplicates were removed. Results from each approach were checked against a “gold standard” (GS) of case reports previously identified in two systematic reviews (75 anti-TNF agent, 57 baclofen case reports).  Sensitivity and precision of each method were calculated. The broadest search, using drug terms alone, yielded over 11,000 and over 3,600 references for each systematic review topic.  Sensitivity decreased as search methods became narrower; precision was consistently low (generally < 6%).  The search method for drug terms alone had 100% sensitivity for both systematic review topics with very low precision (<1.6%).  Precision was highest when drug, disease, case report and AE were combined for baclofen (15%) with low sensitivity (26%). Of the GS articles available from MEDLINE, 7% and 18% of anti-TNF agent and baclofen case reports respectively were not indexed as case reports. Sensitive search methods able to identify relevant case reports are important, but when a sensitive search was constructed, this lead to low precision.  A sensitive search method must be broad, and search both databases. Precision remains low with each combination of approaches, making accurate identification of case reports rather labor intensive. This illustration demonstrated the extent to which decisions made when developing search methods impact the comprehensiveness of reviews. Further work is on-going to confirm the generalizability of these findings to other drugs.