The study examined the individual, job related, and training variables that affected turnover in six NAB-JOBS Federally funded training programs for the disadvantaged. Three different sources of data were used. First, demographic and wage information was obtained on all trainees in five companies in order to run a regression analysis. Second, case studies of the companies were developed through interviews with company officials, program staff, and supervisors in each company. Third, interviews with a sample of retentions and separations from five companies were held to measure attitudinal and motivational differences between the two groups. Wage and individual variables were found to be most highly related to turnover. Males, youth, and blacks had significantly higher turnover. Wages were also shown to be highly related to turnover. There was little evidence indicating that any other single job or training variable had a significant effect on turnover. (Seventeen tables and figures and a bibliography are included.) (LY)