A study compared job-performance-aided (JPA) training developed for a nontechnical area (census enumerator/interviewer) with training developed using the instructional systems design (ISD) model. (JPA training, an elaboration of the ISD approach, requires redesign of job reference materials into job-performance aids to simplify training and job performance.) Three pairs of census district offices, matched on variables related to difficulty of enumeration, were randomly assigned to the two training methods and used training guides prepared using the two models. Data from 1197 JPA-trained and 1389 control enumerators were available for analysis. Attitudinal data, collected before and after job experience, served as basis for comparison. (Only some collected performance data were analyzed for reasons described.) Both training approaches were very successful in terms of positive ratings by trainees. JPA trainees, however, gave significantly higher ratings of training quality, job preparedness, reading ease of training materials, adequacy of training length, and ratings of specific activities. They also reported significantly higher level of on-the-job manual use and better coverage of critical job tasks during training. Supervisors familiar with both training approaches expressed preference for the JPA training. (Questionnaires, job performance record, and data tables are appended.) (YLB)