For observational or prevention studies, accurately identifying by mail and telephone cognitively normal elderly volunteers would be cost effective. We describe how to recruit cognitively normal sib-pairs over age 80 using commercially available lists by age and ZIP code. We mailed an Institutional Review Board-approved letter to 24,366 persons over 85 around Jacksonville, FL, and received approximately 3,000 postcard replies with approximately 500 answering 3 screening statements affirmatively. Of these, we recruited 128 persons who underwent both in-person and telephone evaluations, the latter using the Telephone Interview of Cognitive Status-modified (TICS-m) and Clinical Dementia Rating scale (CDR). Blinded to the TICS-m and CDR data, clinicians made a consensus diagnosis for each participant, 120 were normal and 8 had mild cognitive impairment. With CDR, 119 patients (93%) screened as normal, and of these 115 (97%) were confirmed as normal with the consensus diagnosis. A TICS-m score cut-off of <29 resulted in a similar proportion of normals in the screened sample (97% or 103/106); however, 13 normal volunteers would have been excluded because they scored <29 on the TICS-m. Supplementing the sample, we recruited 12 age-matched cases having consensus diagnosis of dementia (n=2) or mild cognitive impairment (n=10). A CDR>0 correctly identified 12/12, whereas the TICS-m <29 correctly identified 7/12. Hearing loss present in 50% did not influence TICS-m or CDR performance. Using stringent entry criteria and the telephone CDR, this method accurately identified normal elderly persons.