Background Behavioral assessment is useful for the diagnosis of frontotemporal lobar degeneration (FTLD). We explored the ability of the Frontal Behavioral Inventory (FBI) to discriminate between patients with distinct subtypes of FTLD and patients with Alzheimer's disease (AD), as well as the influence of demographic variables on FBI scores. Methods The FBI was administered to the caregivers of 87 patients diagnosed with FTLD [64 behavioral variant FTLD, 19 aphasic variant FTLD (primary progressive aphasia), and 4 motor/extrapyramidal variant (corticobasal syndrome)] and 30 patients with AD. Patients with AD were older than patients with FTLD. The two groups did not differ with respect to duration of illness, level of education, or sex ratio. Results Age significantly predicted disinhibited positive behaviors, such as perseverations and irritability, whereas education did not contribute to FBI ratings. Classification accuracy for the discrimination of AD and mixed FTLD groups was 81%. Moreover, 88.3% and 83.7% accuracy was achieved for the discrimination of AD and behavioral variant FTLD, and AD and primary progressive aphasia groups, respectively. The Total Negative subscale of the FBI, which summarizes the presence of deficit (negative) behaviors, was the best discriminator. A cut‐off score of 17 provided 83% sensitivity and 98% specificity in distinguishing between FTLD and AD patients. Conclusions The FBI is a sensitive and specific tool for the differential diagnosis of FTLD from AD. The optimal cut‐off point for the detection of FTLD patients was lower than that initially proposed. Copyright © 2012 John Wiley & Sons, Ltd.