This study compared the effectiveness of five methods of providing GAL representation: law school clinic (law student with faculty advisor), staff attorney, paid private attorney, lay volunteer with paid attorney (court appointed special advocate, CASA), and lay volunteer (unassisted (CASA). Two examples of each method were examined at nine sites in six States. Data sources included interviews with judges, state's attorneys, and GAL program directors; child welfare agency and court records, and case studies of GAL activities derived from network interviews. The private attorney model was the weakest method of providing representation. Private attorneys generally did not develop independent assessments of the case or conduct adequate investigations, frequently failed to meet with the child before or after court appearances, did not monitor cases, were ineffective in helping the child receive services, and failed to assist in placement decisions. Law students also did not generally perform well in the GAL role because of inexperience with both child welfare and legal systems. The two CASA models clearly excelled in serving the best interests of the child. CASA's were highly rated by professional respondents. They conducted extensive investigations, monitored the case closely for its duration, maintained an independence of viewpoint, developed good relationships with clients, and were most effective in ensuring that the child and family received services that could result in family reunification. Staff attorneys also were effective in serving the child's interests, particularly with respect to legal activities and in services, case goal, and placement outcomes. However, they were weak in followup and maintaining contact with the child. Supplemental data are appended. 11 exhibits, 5 figures, and chapter references.