Research has not provided feasible models to identify dementia in primary care. We construct a broadly based diagnostic algorithm synthesizing information from known risk factors, such as poor cognition, sociodemographic factors, and health history. Data were from the Canadian Study of Health and Aging (CSHA) Phase I. Dementia was diagnosed by clinical consensus. All subjects had a Mini-Mental State Examination (MMSE) score and a Modified MMSE (3MS) score. Multiple logistic regression was used to build our diagnostic algorithm, which was then tested for classification accuracy on the basis of the area under the receiver operating characteristic curve. The area under receiver operating characteristic curve for our diagnostic algorithm using 3MS as a binary variable was significantly greater than the 3MS alone (P<0.001). However, no significant difference was found when using 3MS as a continuous variable in the algorithm. Similarly, a binary MMSE algorithm would provide greater accuracy than MMSE alone. In terms of the usage of our algorithm in practice settings, given the prevalence of dementia, the clear benefits of accurate identification and earlier intervention, adding a few questions to the binary 3MS in our algorithm quantitatively improves the dementia prediction, which is important for patients, caregivers, and health providers.