The Neuropsychiatric Inventory (NPI) is used to assess neuropsychiatric symptoms in dementia patients. To reduce clinicians' time taken to administer the NPI, the authors studied a caregiver-administered NPI (CGA-NPI), in which caregivers completed the written form of the NPI worksheet. After a brief presupervision session, the caregivers of 61 dementia patients completed the CGA-NPI by reading through the worksheet. This was followed by a postsupervision session to check if the caregivers had completed the form appropriately. The correlation between the prevalence rates of each neuropsychiatric symptom obtained by the CGA-NPI and those obtained by the NPI was fair to good (kappa = 0.57-0.90) in all domains. All frequency, severity, and caregivers' distress scores of the CGA-NPI correlated significantly with those of the NPI (r> 0.6, P<.001). Total CGA-NPI scores also correlated highly with total NPI scores (r= 0.86, P<.001). These results suggest that the CGA-NPI can substitute for the NPI, saving administration time.