Integrating science and mathematics in the elementary classroom is a significant challenge, as evidenced by the few convincing examples currently available in the scholarly literature. Yet, this integration is often seen as presenting epistemological and pedagogical value. This study aims to measure the impact of science and mathematics (S&M) learning situations on certain perceptual variables, namely, self-efficacy (SE), interest, and perceived S&M links, by comparing three experimental conditions: (0) control, (1) science, and (2) integration of S&M. Multilevel linear regressions using data from the Science and Mathematics Integration Questionnaire (SMIQ) suggest that making mathematics explicit in science learning situations (2) or keeping mathematics instruction implicit (1) significantly increases SE in science and SE in mathematics compared to the control condition (0). Further analysis suggests that there were significant a posteriori differences between conditions in students’ ability to perceive the links between science and mathematics, with the highest score being for S&M and the lowest for control.