Summary Objectives Basic life support (BLS) by doctors has been shown to be of poor quality. To improve medical education training should be simplified, and simultaneously the learner should be involved more actively. To combine both ideas we trained medical students to give BLS courses and sent them to teach school children. This was a requirement for their emergency medicine course. Our model was compared to conventional teaching. Design Medical students were assigned at random to one of three groups. Group 1 (“university”) attended a conventional university BLS/ALS course. Group 2 (“EMS”) accompanied a BLS vehicle of the emergency medical service (EMS) after suitable preparation. Group 3 (“school”) was instructed to teach BLS and then sent to teach at schools. Main outcome measures Clinically significant BLS skills, and overall BLS skills, each assessed by structured clinical examination (SCE). Theoretical knowledge assessed by written (open question) test. Results Clinically relevant mistakes were seen in 37.5% in group 1 (“university”), compared to 28.8% in group 2 (“EMS”), and 11.3% in group 3 (“school”). Highly significant differences were shown between “school” and “EMS” ( p = 0.011), and between “school” and “university” ( p < 0.001). In practical testing for overall performance the “university” group reached a median of 78.8% (25th–75th percentile 69.2–84.6%), group “EMS” reached 76.9% (69.2–88.5%), and group “school” 84.6% (76.9–90.0). Group “school” showed significant advantages over “university” ( p = 0.015) and “EMS” ( p = 0.010). Written test results did not differ statistically. Conclusion Medical students teaching BLS to school children as a compulsory element of their own medical training showed superior practical skills as compared to conventional teaching. Theoretical knowledge was equivalent to the control groups, although their course contained less theoretical information.