This cross-sectional study examines the association between total prescription medication use and potentially inappropriate medication use (PIRx) among community-dwelling elderly patients with and without dementia. Data (September 2005 to September 2007) were from the National Institute on Aging-funded National Alzheimer's Coordinating Center Uniform Data Set. The study analyzed the Uniform Data Set initial visits of 4518 community-dwelling subjects aged 65 years and above with and without dementia (2665 and 1853, respectively). PIRx was defined using a partial list of the 2003 Beers criteria. Generalized linear mixed models were applied to estimate the association between PIRx and polypharmacy. In both groups (with and without dementia), subjects who received PIRx on average took more medications than those taking no PIRx. As the total number of medications used increased, the odds of having PIRx also increased, controlling for dementia diagnosis and other subject characteristics. Our key findings were consistent after considering 2 definitions of PIRx (with or without oral estrogens) and accounting for missing data. In summary, the total number of medications used is associated with PIRx among Alzheimer's Disease Centers community-dwelling elderly patients with and without dementia, with polypharmacy increasing the risk of PIRx. Ensuring appropriate medication use in this population is clinically important because of the significant risks for institutionalization.