Background: While dyadic faculty-mentored relationship research currently saturates the mentoring literature, recent developments suggest the need for a broader consideration of a student's mentor network. Research taking a network approach may provide deeper insights into the formation and benefits of mentorship for undergraduate students in science, technology, engineering, and mathematics (STEM) disciplines. Utilizing Developmental Mentor Network Theory and ego-centric social network analysis, this pre-registered study evaluates how the characteristics of mentees and mentors relate to both the content of support and structure of mentor networks in a large sample of White and Hispanic/Latino(a) STEM undergraduates across 12 universities. Results: Results were nuanced but showed that perceived psychological similarity with their mentor(s) predicted both dyadic and network average levels of mentor support (i.e., psychosocial, career, role modeling) and relational satisfaction. Furthermore, results point to homophily and engagement in undergraduate research effects on mentor network structures. Conclusions: These findings highlight the importance of using a network approach to deepen our understanding of the factors (e.g., psychological similarity) that may influence the formation and maintenance of robust and diverse supportive mentoring networks.