Background: Until recently, frontotemporal lobar degeneration (FTLD) was considered a rare neurodegenerative disorder that was difficult to diagnose. The publication of consensus criteria for FTLD, however, prompted systematic studies. The criteria categorize FTLD into 3 subgroups: frontotemporal dementia, semantic dementia, and progressive nonfluent aphasia. Objective: To compare demographic characteristics of patients in the 3 FTLD subgroups. Design: We compared diagnostic breakdown, age at onset, sex, Mini-Mental State Examination score at first visit, education, and neuropathological diagnoses in a large sample of FTLD patients from 3 different university dementia clinics, including 2 neurologic clinics in the United States and 1 psychiatric clinic in Germany. Results: The frontotemporal dementia subgroup represented approximately half of all FTLD diagnoses. Patients diagnosed as having frontotemporal dementia (mean age, 57.5 years) and semantic dementia (mean age, 59.3 years) had an earlier age at onset than patients diagnosed as having progressive nonfluent aphasia (mean age, 63.0 years). There were significantly more men diagnosed as having frontotemporal dementia (63.5%) and semantic dementia (66.7%) when compared with progressive nonfluent aphasia (39.1%) (P=.005 for frontotemporal dementia vs progressive nonfluent aphasia and P=.002 for semantic dementia vs progressive nonfluent aphasia). Generally, the demographic features and diagnostic categories of the patient populations across the 3 sites were comparable. There were 68 deaths and 37 autopsies. Frontotemporal lobar degeneration with ubiquitin-positive τ-negative inclusions (48.5%), dementia lacking distinctive histopathological features (18.2%), and Pick disease (15.2%) were the most common neuropathological diagnoses. Conclusions: These findings show that cohorts of patients can be combined using new research criteria for FTLD and demonstrate striking demographic differences among FTLD subgroups. The sex and age-atonset differences suggest that there may be biological differences among FTLD subgroups. In this sample, FTLD with ubiquitin-positive inclusions accounted for half of all neuropathological diagnoses. ©2005 American Medical Association. All rights reserved.