This study investigates the hypothesis of a child differential sensitivity to parenting improvement. One hundred and fourteen parents of preschoolers participated in two parenting micro-trials aiming to increase parental self-efficacy in view of improving child behavior. The first micro-trial took place in a short-term laboratory experiment; the other was an eight-week parenting group intervention, both focusing on altering parental cognition. Differential effects of parental self-efficacy improvement on child's positive and negative behaviors, depending on child temperament, were compared at post-test between control and experimental groups. Both observation and questionnaires were used to measure child behavior as well as regression and Regions of Significance analyses. Child differential sensitivity was found both in the laboratory experiment and in the parenting intervention for the temperamental trait of negative emotionality but not for the temperamental trait of activity. However, this sensitivity was in an unexpected direction. Highly emotional children benefited less from this parental cognitive improvement than children low on emotionality. These results may be explained by the specific cognitive nature of these two parenting micro-trials.