To determine which of several variables best predict aggression in hyperactive children, 38 hyperactive first-, second-, and third-graders were assessed for 12 consecutive school days on 11 behaviors associated with aggression. Behavioral items on the criterion measure, the Daily Behavior Checklist, were summed and then averaged over the 12 days to produce a daily aggression score. Predictor variables included parents' marital conflict, deficits in sustained attention toward children, events during the mother's pregnancy and delivery, parental child-rearing skills, and neuropsychological factors. Daily aggression correlated .36 (13% of variance) with measures of marital and family discord, .40 (16% of variance) with attention performance variables, .57 (33% of variance) with neuropsychological variables, .53 (28% of variance) with prenatal and birth variables, and .54 (29% of variance) with parental variables. Because of the great number of neuropsychological variables, the authors warn that the correlation between those factors and daily aggression be interpreted with caution. The strongest predictor variables from each of the domains were combined in a final regression analysis. Among the strongest individual predictors were the mother's variable level of knowledge of behavioral principles, accidental jarring of the fetus during pregnancy, suspected birth complications, and intensity of the mother's punishment of the child. Preliminary results point to parental child-rearing habits as a potential factor in the existence of aggression in young children. It is suggested that events during pregnancy and delivery have an affect on aggression as well. (Author/RH)