Study data were obtained from documents maintained by Georgia's Department of Labor. The ex-offender sample consisted of 521 individuals aged 21 and over who reported a prior arrest or conviction for a felony as a barrier to employment for the July 1989-June 1990 program year. A random sample of 734 JTPA recipients who were eligible based on barriers to employment or economic disadvantage was also drawn. Employment status was assessed at two time points, termination from JTPA and during followup. Logistic regression analysis was used to analyze employment at termination. Results indicated that ex-offender status had no influence on employment at termination or during followup. However, individuals who were unemployed for 15 or more weeks prior to entry were less likely to have a job at termination and during followup. Participants who received employer-based training were more likely to have a job at termination and during followup. Employment subsidized by JTPA allowed participants to demonstrate their skills and gave employers an opportunity to observe participant abilities in the workplace. The finding that remedial education was negatively related to employment, compared to job search assistance, suggested that educational improvements alone may not be sufficient to improve a trainee's employment chances, at least in the short-term, and that both specific job-related skills and educational growth may be necessary. 34 references, 8 notes, 3 tables, and 1 figure