Theory and empirical evidence indicate that the quality of relationships between mentors and youth is critical to determining the effects that mentoring programs have on youth participants. However, studies of mentoring programs have relied almost exclusively on self-reports of the quality of the mentoring relationship. The goals of the current paper are to discuss the limitations of exclusively relying on self-reports to measure relationship quality, argue for the necessity of incorporating naturalistic observations into measurement, and propose a specific framework for naturalistic observation and rating of these relationships.