Fifty-nine inmates, men and women, from two correctional institutions were randomly assigned to eight groups to test the effect of book discussion on attitudes. The four experimental groups read and discussed weekly a series of six titles during the 12-week program. The four control groups met three times to participate in a reading interest survey. Each leader team, composed of two librarians, led one experimental and one control group. Pretest and posttest scores on Socialization Scale of the "Personal Values Abstract" and on a "Semantic Differential" test of attitudes related to persons and behaviors were subjected to analysis of covariance and the F test. Covariance analyses revealed that the experimental groups were less accepting of Dope Addiction and Stealing, the two behavioral concepts, than were the control groups, while no significant difference was found on attitudes toward concepts relating to persons. Analysis of interactions showed that those in the experimental groups who had served more time, had more time to serve, or were Black were affected more positively by bibliotherapy than their fellows. Conclusions: For those inmates who wish to participate and can read and comprehend, the group discussion form of bibliotherapy may supplement the correctional program: (1) by improving attitudes related to behavior for all groups, (2) by additionally improving attitudes related to persons for certain groups, and (3) when conducted by librarians working with small inmate groups. (Author)