The follow-up of participants in a longitudinal survey can be difficult and expensive. This paper reviews the experiences of Project TALENT's 11-year follow-up survey of 400,000 individuals tested as high school students in 1960 and the experiences of follow-ups in other longitudinal studies. Methods for minimizing sample attrition and detecting and correcting for nonresponse bias are discussed. A description of the methods used to locate mail survey nonrespondents is included along with analysis of the cost effectiveness of each method. The results of the efforts to reduce attrition and correct for nonresponse bias are analyzed in terms of a measure of general academic aptitude and an index of socioeconomic status. (Author)