The Child Abuse Potential Inventory was completed by 56 Head Start staff. Because the Head Start program creates an environment where staff work almost exclusively with parents or with children, it was possible to examine the perceived differences among workers attracted to child or parent work. Scores for classroom and non-classroom staff were compiled and examined by analysis of variance for significant differences. The results revealed that, while the mean abuse score for classroom staff was approximately 17 points higher than the mean score for non-classroom staff, both scores were within the normal, nonabusive range and the difference was not statistically significant. Stepwise regression was used to develop a prediction equation for potential abuse scores. Variables used in the analysis were age, race, marital status, number of children in the home, education level, and staff position. Race was the only variable that entered into the equation. The implications for use of such information in the screening, hiring, and assignment of staff to various work roles and responsibilities are explored. References are included. (TE)