Intrapersonal split alliances were defined as discrepancies in how group members perceived their positive bonding relationships with the group leader, the other group members, and the group-as-a-whole, and were related to group members’ outcome. Participants were 168 patients (116 women and 52 men) diagnosed as overweight or obese who participated in 1 of 20, 12-session, therapy groups for weight management. Group members completed the Outcome Questionnaire-45 (OQ-45, Lambert et al., 2004) pre- and posttreatment and the Group Questionnaire (GQ, Krogel et al., 2013) at early, middle and late group sessions. Early, middle, and late ratings were aggregated because bond scores were consistent across time. Two-level (members within groups), polynomial regressions and response surface analyses were used to examine congruency and discrepancy in ratings of Positive Bonding Relationships to the leader, group members, and group-as-a-whole. When the discrepancy between positive bonding relationships with the leader and positive bonding relationships with the group members increased, and when the discrepancy between positive bonding relationships with the group members and positive bonding relationships with the group-as-a-whole increased, there was less symptom improvement. The findings show that, like interpersonal split alliances, intrapersonal split alliances are harmful for treatment progress and need to be recognized and addressed by the group leader.