Objectives: Telephone screening has become a common method used in health services research to identify efficiently persons in specific populations of interest. In this research, we used a large-scale telephone screening survey to assess: (1) the effectiveness of the telephone method in gathering tooth count information by measuring response rate (cooperation) to specific questions and (2) the validity of subjects' reports of the number of remaining natural teeth. Methods: We used a telephone screening methodology to identify dentate persons (at least one natural tooth remaining) who were 45 years old or older and resided in one of four counties of north Florida. At a second stage, a sample of the telephone screening participants was selected for further study, which consisted of a baseline in-person interview and a clinical examination. We compared the number of remaining teeth reported during the telephone interview with the number determined at baseline examination. Results: The telephone method was effective at gathering tooth count information, although response rates varied with the level of specificity required. Almost all subjects reported the number of teeth at least at the nominal and ordinal levels, but fewer than three-fourths reported the number at the interval level. When the unit of analysis was the overall sample, self-reported number of teeth was a valid measure of the true number. When the unit of analysis was the individual subject, validity was associated with certain clinical and sociodemographic factors. Conclusions: When the unit of analysis is the overall sample, these results suggest that self-reported tooth counts during a telephone interview are sufficiently valid to meet all but the most stringent data requirements. When the unit of analysis is the individual subject, these tooth counts may not be valid, depending upon the degree of specificity required and subject characteristics.