Ongoing changes in available agents and health care delivery systems have made it imperative to study the quality of antidepressant use in vulnerable and traditionally underserved elderly. We conducted a retrospective cohort study among 12,130 new antidepressant users aged ≥65 years with a recent diagnosis of depression in the Pennsylvania Pharmaceutical Assistance Contract for the Elderly Program from January 1, 1994, to December 31, 1999. Additional use information was available through Medicare data. Potentially hazardous antidepressant regimens were defined as use of highly anticholinergic agents or daily dosages in excess of geriatric prescribing guidelines. Low-intensity regimens were defined by lower than recommended daily dosages, too-short durations of therapy, or lack of follow-up. Of all elderly antidepressant users, 43.3% were taking suboptimal regimens. Potentially hazardous regimens were used by 11.9%, including 7.3% taking highly anticholinergic agents and 5.3% using excessively high daily dosages. Low-intensity regimens were used by 34.8% of patients, including 7.6% with excessively low daily dosages, 19.3% with short durations of therapy, and 14.8% with inadequate follow-up. Potentially hazardous regimens were associated with ages 65 to 74 years, nursing home residence, cancer diagnoses, less comorbidity, use of other psychiatric medications, making more physician visits, and earlier calendar years. Low-intensity regimens were associated with ages ≥85 years, nonwhite race, greater comorbidity, fewer physician visits or inpatient days in the baseline 6 months, and not using other psychiatric medications. Suboptimal antidepressant use remains common in the elderly, especially the use of inadequately intensive regimens. Interventions are needed to improve the quality and outcomes of antidepressant use in this vulnerable population.