Twenty primary school students were randomly assigned by a matching procedure to either an experimental or control group. During the whole school year, both groups were taught mathematics according to the curriculum in the school; however, the groups followed different programs in fractions. Students in the experimental curriculum were provided with opportunities to develop number sense in a program where class discussion and the number line takes a central place as a mental model for formal reasoning in learning fractions. By contrast, in the curriculum of the control group the circle was used as a model to reach formalization. At various time points, the teaching and learning processes and the outcomes of the two curricula were investigated by participant-observations, in-depth interviews and tests for mathematical understanding. This comparison particularly shows qualitative differences in operating with fractions between experimental group and control group. In their mental models, strategies and reflections students from the experimental group demonstrated clearly more insight i.e., aspects of number sense in processing fractions, both relatively when compared to the controls, as absolutely in regard to the national standards. Moreover, we observed differential effects in comparing the experimental with the control group. The outcomes of the study are discussed and some implications for curriculum theory and practice are presented. (Contains 65 references. (Author/CCM)