The NBER-Mathematica Survey of Inner-City Black Youth was intended to be a random sample of youths from low-income, predominantly black areas of three cities.'Yet older youths are poorly represented in the survey, with half as many in their twenties as in their teens. Comparisons of population counts between the NBER and 1980 census data for targeted areas suggest that this difference is the result of the NBER survey having undercounted older youths. This appendix examines why the undercount may have occurred and evaluates its possible biasing of the results presented in this volume.