This study investigated expectations of social rejection in adolescents with anorexia nervosa. Female adolescents (N = 24) admitted for hospital treatment completed both a positive interpretation bias training (experimental training) and a 50% positive and 50% negative interpretation bias training (control training), in a counterbalanced order, in two different testing sessions using a within-subjects design. The primary outcome for the training was the frequency of negative interpretations produced in the sentence completion task. At baseline, participants produced more negative than benign interpretations of ambiguous social scenarios (medium/large effect size). Following experimental training, participants produced fewer negative and more benign interpretations compared to control training (medium/large effect sizes). A trend for higher levels of self-esteem following virtual ostracism was found in the experimental condition compared to the control condition (small/ medium effect size). Interpretation bias modification has the potential to improve resilience to rejection in adolescents with anorexia nervosa.