Gender differences in achievement are often explained by behavioural differences between girls and boys or by differences in the way the social environment stimulates or limits students to do well in certain subjects. These studies seem to have one thing in common; the underlying assumption that gender differences in achievement are (partly) the result of the influence of affective factors, such as attitudes or self-confidence. Multilevel analyses on the Dutch data of TIMSS-2003, showed that among grade 4 students with a similar level of self-confidence in mathematics, the girls’ scores were higher than those of boys (Meelissen & Luyten, in press). The reason for conducting this Dutch study was the higher achievement of boys compared to girls on the TIMSS-test. Just as in most other countries, this situation is reversed for reading literacy: boys are outperformed by girls on the PIRLS-reading test (Mullis, et al., 2006). The question occurs how important self-confidence and other affective factors are for girls and boys when reading literacy is concerned. For this paper, the analyses on the relation between gender, affective factors, and achievement in mathematics (TIMSS-2003) are repeated for reading literacy (PIRLS 2006). The problem statement of this paper is: To what extent are subject related affective factors, predictors for mathematics achievement and reading achievement of girls and boys, in grade 4 primary education? Analyses The study consists of four steps. First, the reliability and validity of the available student, class, and school characteristics in PIRLS are explored. Next, Pearson product-moment correlations between student characteristics and achievement are conducted in order to select student factors for the final analyses in PIRLS-data. The third step consists of multilevel analyses with MlwiN. Finally, the results of the explorations on TIMSS and PIRLS are compared. This comparison is descriptive, taking the differences in operationalization of the factors in the school, teacher and student questionnaires of the two studies into account.