To compare detection of cognitive impairment using the Mini-Cog and Mini-Mental State Examination (MMSE) and to identify sociodemographic variables that influence detection in an ethnoculturally diverse sample. Cross-sectional. A registry of the University of Washington Alzheimer's Disease Research Center Satellite. A heterogeneous community sample (n=371) of predominantly ethnic minority elderly assessed using a standardized research protocol, 231 of whom met criteria for dementia or mild cognitive impairment (MCI). Demographic data, a standardized research protocol for cognitive assessment and dementia diagnosis, MMSE, and Mini-Cog. Both screens effectively detected cognitive impairment, the Mini-Cog slightly better than the MMSE (P<.01). Overall accuracy of classification was 83% for the Mini-Cog and 81% for the MMSE. The Mini-Cog was superior in recognizing patients with Alzheimer-type dementias (P=.05). Low education negatively affected detection using the MMSE (P<.001), whereas education did not affect the Mini-Cog, and low literacy minimally affected it. The Mini-Cog detects clinically significant cognitive impairment as well as or better than the MMSE in multiethnic elderly individuals, is easier to administer to non-English speakers, and is less biased by low education and literacy.