Dementia screening measures serve an important role in epidemiological research, clinical trials, identifying patients for more comprehensive assessment, and monitoring progression. Telephone-administered measures allow increased flexibility. Although there are several extant telephone-administered measures that offer the promise of efficient detection of cognitive impairment, research evidence supporting their use is limited. We review telephone-based cognitive screening instruments for detecting dementia or mild cognitive impairment, critically review the evidence for their validity, and make recommendations for future research directions. Most measures reviewed do a good job of classifying patients as likely to have dementia or likely to be neurologically healthy. However, with sensitivity values ranging from 38% to 100% and specificity values ranging from 79% to 100%, there is considerable variability in the measures available. Future validation studies should reflect the populations most likely to benefit from a telephone-based measure, namely community-dwelling elderly who have not yet been identified as being cognitively impaired.