Graduate students in biomedical graduate programs can have negative experiences with their research advisors, which lead to worse well-being and lower success. Further, conflicts with advisors can be worse for biomedical graduate students from underrepresented backgrounds (e.g., racial and ethnic minorities). Few intervention programs have examined how to improve student-advisor relationships in an ongoing, sustainable way. The IMPACT Study: Improving Mentorship Practice through Attributions and Conflict Training aims to develop and test the effectiveness of a research advisor mentorship intervention with two complementary components: (1) attribution retraining, to encourage advisors to perceive their mentorship relationships as controllable and malleable in order to motivate them to improve their mentorship, and (2) conflict skills training, to equip advisors with specific and actionable ways to manage and resolve the inevitable conflicts that arise in these enduring relationships. A second goal is to examine whether a mentorship training program can be successful and sustainable with modest time and resource investments, to facilitate scale-up.