In many music teacher education programs improvisation remains undervalued, even though it is often cited as essential for musical fluency. Many pre-service music teachers feel underprepared to teach it. This study addresses this gap by modeling how five constructs—attitude to studying improvisation (AS), attitude to teaching improvisation (AT), attitude to including improvisation (AI), self-efficacy for improvisation (SEI), and self-efficacy for teaching improvisation (SETI)—interrelate to influence teaching practice. We collected survey data from 123 pre-service music teachers across ten Dutch conservatories and universities for applied science and conducted semi-structured interviews with 21 pre-service music teachers and 10 music teacher educators. We used PLS-SEM (partial least squares structural equation modeling) to test hypothesized paths among AS, AT, AI, SEI, and SETI.The PLS-SEM results indicated that SEI was the strongest direct predictor of SETI. AS had a significant effect on SEI and influenced SETI only indirectly through SEI. AT has both a direct effect on SETI and an indirect effect mediated by SEI. AI, despite yielding the highest mean scores, did not significantly predict either SEI or SETI. Qualitative data illuminated how mastery experiences, proper challenges, safe environments, peer feedback, and scaffolded practice foster SEI, which then supports SETI. Age and gender differences emerged: older participants often reported stronger AS but weaker SETI; female participants tended to report lower SEI and SETI.Overall, SEI serves as the key variable linking attitudes to teaching confidence. To better prepare pre-service music teachers to teach improvisation, teacher education programs should prioritize strengthening personal improvisation competence, then offer scaffolded teaching experiences with feedback, while actively addressing equity in confidence development.