Thirty mothers of Title 1 children (5-11) were matched for marital status, sex of children, ethnic background, age, and level of formal education. They were randomly assigned to a waiting-list control group, a parent treatment group, and a parent-and-significant-others treatment group. The dependent measures were the following: (a) "the way I feel about myself," (b) parent-child relations, (c) child-parent relations, (d) Parental Attitudes Test, (e) Miskimins Self-Goal-Other Discrepancy Scale, (f) "behaviors you like," (g) "behaviors you dislike," (h) parent evaluation of the program, (i) Metropolitan Achievement (Readiness) Test, and (j) school grades. Data were collected from four people: the mother, the targeted child, a sibling, and the teacher. Both treatments were significantly effective and equally effective in producing and maintaining positive gains in the targeted behaviors. Both treatments were significantly more effective than the control condition in producing positive generalization effects across behaviors, across settings, across time, across response systems, and across persons. Although there were no consistent and robust differential effects among the two treatments, the generalization effects were particularly striking since such effects have been characterized as exceptions rather than the rule. (Author)