Every clinician faces the daily question of which antidepressant is best for a particular depressed patient. Double-blind studies submitted for U.S. Federal Drug Administration marketing approval include only the "purest" population of patients, and the American Psychiatric Association and other treatment guidelines often do not adequately address the complexities of developmental, family history, psychosocial, medical, and psychiatric comorbidity, and treatment-refractory issues that are seen in routine clinical practice. Long-term trends in depression treatment include ever-expanding choices among drugs, highly specific psychotherapies, and attempts to treat chronic and/or mild cases, with the goal of remission for all patients. We performed literature reviews and attempted to synthesize factors that may be useful in the application of evidence-based medicine in office-based psychiatric practice. We have found that factors influencing antidepressant selection include drug factors (including tolerability, interactions, and cost), depression subtype, psychiatric and medical comorbidity, and stage of life. In addition, patient preference for avoiding certain side effects and personal and family history of treatment response are helpful information. Most patients in the community would not fit strict antidepressant study criteria. Biologic markers predicting treatment response are not yet widely available, so the optimal choice of medication must be guided by detailed history.