The paper describes a variety of analytical difficulties facing prospective users of the first follow-up of the National Center for Education Statistics National Longitudinal Survey of the High School Class of 1972 (NLS) and suggests some possible approaches to coping with these. The primary focus is on the causes and consequences of selective item non-response in the first follow-up survey. Coding schemes used to flag this non-response and alternative approaches to estimating values for missing data are discussed. An examination of the special codes used for routing-pattern errors and missing data leads to the proposal of preparation of an analysis-oriented data file to parallel, but not replace, the existing documentary file. Certain coding modifications are mentioned which might be implemented for such a file. An examination of patterns of item non-response leads to the conclusion that the questionnaire's content and format, especially requests for detailed and/or private information, complex routing patterns, and a layout better suited to personal interviews than to mail-out collection, are probably responsible for some item non-response. Possible modifications that might reduce item non-response in future follow-up surveys are suggested. Review of several approaches to adjustment for missing data leads the authors to recommend a specific imputation procedure for data already collected. Also described are some possible methodological studies aimed at testing the effects of data assignments upon characteristics of the present NLS data base. (Author/RC)