The target population identified for the purposes of this survey was all DCWs who were part of the day or evening shift (n=172) in two regional detention centers that serve two large metropolitan areas and counties in a Southeastern State. Self- administered questionnaires were distributed at each center on several different occasions during the months of September and October 1991 to groups of DCWs who volunteered to participate and were relieved from their shift for the time required to complete the survey. A total of 109 workers from both centers responded for an overall response rate of 63 percent and rates of 71 percent and 50 percent from the two centers, respectively. Likert-type scale questionnaire items that appear to reflect a restrictive or punitive orientation and a general view that detained youth required a more controlling response were combined to create a Punishment/Control index that yielded a reliability coefficient of .71. Several demographic items included in the questionnaire -- race, gender, age, and educational attainment -- were chosen for analysis in this study. Regarding professional/occupational characteristics, the questionnaire included several structural or objective measures, including work schedule, rank, and tenure, as well as more subjective or experiential indicators of variation in the job situation of these workers, such as job stress and role conflict. The study found that demographic indicators (i.e., age and gender) and organizational environment contribute disproportionately to explained variance in punitive orientation once other variables are controlled. Young male DCWs were more likely to have punitive attitudes toward juvenile detainees; however, punitive orientation is apparently primarily a function of organizational and structural influences. Implications for detention policy, staff recruitment, and training are discussed. 4 tables, 13 notes, 39 references, and appended list of individual items on job stress index and individual items on rehabilitative motivation index