In the context of a randomized clinical trial of psychoanalytic psychotherapy (PPT) versus cognitive behavior therapy (CBT) for bulimia nervosa (BN), this study performed secondary analyses of (a) the relation between attachment and pretreatment symptom levels, (b) whether client pretreatment attachment moderated treatment outcome, (c) whether change in client attachment was associated with symptomatic change, and (d) whether client attachment changed differently in the 2 treatments. Sixty-nine women and 1 man of a mean age of 25.8 years diagnosed with BN were randomly assigned to either 2 years of weekly PPT or 5 months of CBT. Assessments at intake, after 5 months, and after 2 years included the Eating Disorder Examination to assess eating disorder symptoms, the Adult Attachment Interview to assess client attachment, and the Symptom Checklist 90-R to assess general psychiatric distress. Repeated measures were analyzed using multilevel analysis. Higher scores on attachment insecurity and attachment preoccupation were associated with more frequent binging pretreatment. Pretreatment attachment did not predict treatment outcome. In PPT, but not in CBT, reduction of binging was associated with an increase in attachment security. The 2 treatment types were not associated with significantly different patterns of attachment-related change. Degree and type of attachment insecurity is related to the frequency of binging in BN. Increase in attachment security may be a treatment-specific mechanism of change in PPT for BN.