A subject search, or a controlled language search, uses pre-defined vocabulary developed by each database.   As an item like a book or journal article is indexed and abstracted, subject headings based on the database's controlled vocabulary are assigned to indicate the main content of the item.  Each item may be assigned anywhere from two or three subject headings to around seven or eight depending on the database. This search strategy is an excellent way to retrieve highly relevant information. For this strategy to be effective, however, you must use the appropriate term, which can be found in the database's online thesaurus.  Because controlled vocabulary is database specific, you should consult the database's thesaurus as you begin searching.  OVID databases like Medline employ a strategy called "mapping" which leads users from an incorrect term to a correct one or from an imprecise term to a precise one.  In this case, consulting the thesaurus isn't quite as important.

A keyword or word search, also called a free-text search, relies on natural language rather than controlled vocabulary.   This type of search most commonly retrieves records which contain the given word(s) in the title, subject, or content notes, though some databases may include other fields in the search.  You will retrieve more entries or hits using  this strategy, but it can be imprecise and can result in a high number of irrelevant hits.

A common search strategy is to start with a subject search with help from the database thesaurus and then, if you aren't retrieving any or enough items, try a word search .  Beware of a search that doesn't produce any results if you know the topic has been written about.  Most often, it's an indication that a problem exists with your search strategy.