Biosocial influences on sex differences, found for school achievement test results in grades 3 and 6, have been studied by means of opposite-sex twin pairs and singleton controls, attending the same classes as the twins. As expected the opposite-sex twin pairs tend to be more similar in achievement test results in Swedish and mathematics than their controls. This is, however, more pronounced for Swedish and could probably be due to mathematics being more susceptible to the type of training experienced at school, while achievement in the Swedish subject is more dependent upon both home background and heredity-environment interaction. Both the twins and their controls tend to become less similar within pairs in achievement test results from grade 3 to 6. It does not seem, thus, as if the educational impact experienced by twins or non-twins, attending the same classes, will make them more similar in this respect. It is also evident that the increasing sex differentiation between grades is not primarily due to home background influences, but rather can be explained both by differential educational treatment and differentiation of abilities with increasing age. (Author)