Mentoring has a mystique that only good can come from it, that at the worst, mentoring programs will not accomplish all that they could, but the youth served will at least be better for the experience. Both impact and process evaluations are needed to answer questions about the real benefits and any potential adverse effects of mentoring. The recent evaluation by the Manpower Demonstration Research Corporation of the national Career Beginnings program does indicate beneficial effects on educational aspirations and college attendance for average students who were not at very high risk. Project RAISE, in Baltimore (Maryland), a program to reduce the dropout rate of high-risk students, provides another example of program evaluation (conducted by McPartland and Nettles, 1991) that demonstrates that school attendance is a behavior that mentors can influence positively. Studies of the process of mentoring have suggested that the right kind of receptive youth can be benefitted by the right kind of motivated and trained mentors. However, research has generally not accounted for the impact of how a program is implemented. Research needs to investigate program implementation and the fit between the mentor and the youth. Good evaluations can help program managers understand and replicate this fit. Information collected about the mentor and the youth can highlight the relationship, participants' perceptions and assessments of the relationship, and obstacles and problems that stand in the way of a gratifying and useful relationship. Evaluation of mentoring programs needs to concentrate on the relationship between the mentor and the youth. (SLD)