Although formative assessment is regarded as a promising way to improve teaching and learning, there is considerable need for research on precisely how it influences student learning. In this study we developed and implemented a formative assessment intervention for mathematics instruction and investigated whether it had effects on students' interest and achievement directly and via students' perception of the usefulness of the feedback and their self-efficacy. We conducted a cluster randomized field trial with pretest and posttest. The 26 participating classes were randomly assigned to a control group or the intervention group. Results of path analyses indicate that feedback was perceived as more useful in the formative assessment condition, self-efficacy was greater, and interest tended to increase; learning progress did not differ between the groups. The assumed indirect effects were partly confirmed: formative assessment showed an indirect effect on interest via its perceived usefulness.