Background: Twin studies of childhood behavior problems support the conclusion that individual differences in impulsivity, hyperactivity, and inattention are largely due to genetic influences. Non-genetic variation is due to environmental influences that are unique to the individual, and possibly to rater contrast effects. In the present longitudinal twin study, we report on the size of genetic and environmental effects on individual differences in attention problems at ages 3, 7, 10 and 12 years. Methods: Mothers were asked to complete the CBCL for their twin offspring when the children were 3 (n = 11,938), 7 (n = 10,657), 10 (n = 6,192), and 12 years old (n = 3,124). We focus on the Overactivity (OA) scale in the Child Behavior Checklist (CBCL/2-3), and on the Attention Problem (AP) scale of the CBCL/4-18. The data were analyzed using longitudinal structural equation modeling. Results: Broad heritability of OA and AP is estimated at nearly 75%, at each age. A contrast effect was observed at age 3 only. The results revealed less stability of OA at age 3 to AP at age 7 (r = 0.40), compared to the stability from AP at age 7 and beyond (r = 0.70). Genetic effects explained between 76% and 92% of the covariance between OA and AP. Conclusions: OA and AP are highly heritable at all ages in both genders. The same set of genes appears to be expressed in boys and girls. The size of genetic and environmental contributions remains the same across the ages studied. Stability in OA and AP is accounted for by genetic influences. Children who do not display OA or AP at a given age are unlikely to develop these problems at a subsequent age.