Longitudinal studies suggest that naturally occurring mentoring relationships are associated with positive, long-term educational outcomes, but little is known about the mechanisms underlying these potential effects. Predominant theories and research methods portray natural mentoring as a static, one-to-one phenomenon and are removed from specific sociocultural and developmental contexts. Thus, the current paper critically reviews existing literature and proposes a dynamic, network-based process model of mentoring. Building on Rhodes' (2005, Handbook of youth mentoring, Thousand Oaks, CA: Sage, 30) theoretical model, the proposed model integrates sociological, educational, and psychological research to examine developmental pathways through which natural mentoring relationships might promote underrepresented (e.g., low-income, ethnic/racial minorities) students' educational persistence and success. In particular, mentoring relationships might enhance students' positive help-seeking beliefs and trust in nonparent adults. These mentors may possess key cultural capital and knowledge that help underrepresented students navigate White, upper- or middle-class cultural norms. Mentors may also cultivate students' school-relevant selves and bicultural identities, facilitating integration into mainstream academia while affirming cultures of origin. Through these processes, previous mentoring experiences may promote the formation of new mentoring relationships, enabling youth to build dynamic support networks. Unfortunately, on a larger scale, advantage and disadvantage persist due to inequitable access to mentoring in educational settings.