OBJECTIVE: To investigate which neuropsychological tests can discriminate between behavioral variant frontotemporal dementia (bvFTD) and psychiatric disorders presenting with similar late-onset frontal behavioral changes, such as apathy, disinhibition, reduced empathy, or compulsive behavior. METHODS: Patients presenting with frontal behavioral changes in middle or late adulthood received extensive baseline examinations, including neuropsychological assessment and brain imaging. After 2 years, examinations were repeated and patients were diagnosed according to DSM-IV or international bvFTD consensus criteria. The study period was April 2011-June 2015. Two groups were selected: 32 patients with bvFTD and 53 patients with a psychiatric or psychological diagnosis. Associations between neuropsychological test scores and diagnostic group were investigated with logistic regression analyses, and diagnostic accuracy was investigated with a receiver operating characteristic curve. RESULTS: BvFTD patients scored lower on tests for confrontational naming, gestalt completion, and verbal abstraction compared to psychiatric patients (P < .01). The confrontational naming test (Boston Naming Test) showed the strongest association with diagnostic group: a lower score indicated a higher probability for a bvFTD diagnosis (P < .001). This test could discriminate between the groups with good diagnostic accuracy (area under the curve = 0.81). Tests for attention, memory, and executive functions showed no discriminative ability between the groups. CONCLUSIONS: Although one of the criteria of bvFTD is low performance on executive tests, these tests are not useful in differentiating bvFTD from psychiatric disorders. We recommend administering language tests, especially an extensive confrontational naming test, to aid differentiation between bvFTD and a psychiatric disorder in patients presenting with late-onset frontal behavioral changes.