The authors analyzed clinical research articles (1980 to 1982) in four general medical journals: New England Journal of Medicine, Annals of Internal Medicine, Journal of the American Medical Association, and British Medical Journal. Of the 1943 articles surveyed, 38.2% were judged to be clinically "relevant" to the elderly. "Relevant" articles were further analyzed to determine whether they contained enough older subjects to enable the reader to draw any inferences about the particular disease or diagnostic or treatment modality studied. Criteria for "age sufficiency" included: mean age of study group (means) greater than or equal to 60 years; if means was less than 60 years, two SD or the range included at least one patient aged 70 years; and greater than or equal to 20% of the study group was older than 60 years. Overall, slightly more than 50% of clinically relevant articles had too few older subjects by the authors' quite liberal age criteria. No statistically significant differences were found among the four journals. Of the age-relevant articles, 25.5% had a mean age of sample studied of greater than or equal to 60 years, 10.9% of greater than or equal to 65 years, 5.6% of greater than or equal to 70 years, and 2.3% of greater than or equal to 75 years. The average age of all subjects studied for three conditions commonly found among the elderly was calculated: coronary artery disease (51.7 years); hypertension (49.8 years); and cancer (52.9 years). The authors concluded that many of the original articles of these four journals did not include (for 1980 to 1982) enough older subjects in their study samples to enable readers to draw any valid inferences about the older population itself.