International comparisons reveal that lower-secondary-level students in Austria perform below the OECD mean in science. Guided by the search for remedies and improvements in science teaching, this study investigates whether cross-age peer tutoring is an appropriate method for teaching physics. A modern and concise definition of peer tutoring is available from the review of previous findings which focus on tutors as well as on tutees. This clarification leads to an experimental setup mostly in tutor-tutee dyads where a sample of n?=?172 students from grades 5 to 8 underwent a cross-age peer tutoring process dealing with different topics within the context of electricity. The overall achievement in electricity for this age group was examined in a pretest-posttest design, using test items about electricity. Additionally, analyses were carried out in order to investigate whether or not there is a correlation between the possible roles within the process (active tutors-passive tutees) and the overall achievement. The results indicate that the active role is a crucial one for the achievement. Finally, a multiple linear regression model is presented which summarizes the research results and estimates the posttest scores based on the relevant parameters: the pretest score, the active role within the tutoring process, and the first language.