Study subjects were male adult inmates, both first offenders and recidivists, housed in two minimum-security institutions. The inmates were pretested for anxiety levels and assigned to 12 different 8-week community counseling groups operated over a 10-month period by volunteer workers. The workers were grouped according to interpersonal skill levels, based on the Group Assessment of Interpersonal Traits (GAIT) technique for the systematic selection of psychotherapeutic talent tests. Several personality tests were also administered to the volunteers and to the inmates with whom they would be working. Small groups consisting of equal numbers of prisoners and volunteers held focused discussions on topics concerning the law and law violators. The effectiveness of the interactions between volunteer counselors and inmates and the effects of volunteers' interpersonal skills were measured by different techniques, including the Interaction Process Analysis and the Group Centered Counseling Scale. Findings indicated successful interaction between high functioning volunteers and nonanxious prisoners, whose anticriminal attitudes were signficantly improved by the close relationships with the workers and by the challenging and friendly confrontations that took place within the counseling groups. Nevertheless, high functioning volunteers did not interact as successfully with anxious prisoners: the outreaching, challenging techniques of the same workers who had favorably influenced the nonanxious inmates acted as a stressor upon the anxious inmates and inhibited rather than promote anticriminal attitudinal gains. Results generally provide additional support for use of the community group format for examining significance of differential association principles; for the less anxious offenders, the structural manipulation of the intimacy factor was associated with anticriminal attitudinal gains. Three tables and references are provided.