The following conclusions may be drawn from this article: 1. The age at which subjects are considered "elderly" or "geriatric" varies from study to study. By convention, age 65 is usually taken as the minimum age, yet virtually all studies cited included subjects below 65. Indeed, in a majority of studies, most of the subjects were below the conventional age of 65. Defining the minimum age for geriatric psychopharmacology studies is particularly important because there may be considerable pharmacodynamic and pharmacokinetic heterogeneity in people aged 65 and older. Clinical experience suggests that patients below age 80 tend to respond to medications like late middle-age adults, whereas those above age 80, as a group, are more sensitive to drugs. Because people 80 and above constitute the fastest growing group of Americans, it is essential to conduct more therapeutic clinical trials using them as research subjects. Information gathered from persons aged 60 to 79 may not be relevant to this older age group. 2. The total number of individuals studied in carefully controlled, double-blind research studies of MAOIs, atypical antidepressants, or psychomotor stimulants are relatively few when compared with the number of individuals studied with tricyclic antidepressants, or the number of nonelderly subjects studied. 3. Although many older persons are included in mixed-age research populations, the number of studies specifically designed in the elderly is also remarkably few. 4. In those studies that did focus on older patients, the definition of elderly was extended "downward" to include patients at age 55, or even 50. Whether or not this was done in order to enhance research subject recruitment, the net result is to vitiate any overall conclusion regarding the efficacy of drugs in patients over age 65, the usual lower cutoff for "elderly" or "geriatric" definition. 5. With the exception of a handful of patients receiving fluoxetine and methylphenidate, virtually no research patient above the age of 80 has been studied using any of the compounds discussed in this article. 6. Some of the double-blind studies reviewed are more than a decade old and techniques in research design have changed substantially since those studies were conducted. Many of the earlier studies would not be acceptable according to contemporary research design criteria because of inappropriate or inadequate inclusion criteria, inappropriate or inadequate outcome criteria, and inappropriate or inaccurate diagnostic criteria.(ABSTRACT TRUNCATED AT 400 WORDS)