It is most helpful to articulate a clear, specific, and concise statement of your research question.  This will help you identify criteria for selecting relevant articles out of a list of search results.  A useful framework to help you develop your research question the PICOTS format.

P (Patients, Population)

P (Problem)

I (Intervention)

C (Comparator(s))

O (Outcome(s))

T (Time to follow-up of each outcome)

T (Timeframe of literature search)

S (Setting)

S (Study design)

Good, specific example:  Compared to hormone replacement therapy, how effective is raloxifene in preventing osteoporotic fractures in postmenopausal women in the first year of therapy?>

Unfocused example:  Is raloxifene good for older women?

If you are not sure of the amount of research available on the topic of your search question, you may want to start with searching on the P (problem) and I (intervention) criteria initially.  If there are many results, you may then start adding additional criteria from your PICOTS question to your search.

Choose databases to search.  Typical examples are PubMed and EMBASE for biomedical and pharmacological topics.  Other disciplinary databases and multidisciplinary databases are available.