Background: Cognitive-behavioural therapy (CBT) has proven to be effective for anxiety-based school refusal, but it is still unknown how CBT for school refusal works, or through which mechanisms. Aims: Innovative statistical approaches for analyzing small uncontrolled samples were used to investigate the role of self-efficacy in mediating CBT outcomes for anxiety-based school refusal. Method: Participants were 19 adolescents (12 to 17 years) who completed a manual-based cognitive-behavioural treatment. Primary outcomes (school attendance; school-related fear; anxiety) and secondary outcomes (depression; internalizing problems) were assessed at post-treatment and 2-month follow-up. Results: Post-treatment increases in school attendance and decreases in fear about attending school the next day were found to be mediated by self-efficacy. Mediating effects were not observed at 2-month follow-up. Conclusions: These findings provide partial support for the role of self-efficacy in mediating the outcome of CBT for school refusal. They contribute to a small body of literature suggesting that cognitive change enhances CBT outcomes for young people with internalizing problems. Regarding methodology, the product of coefficient test appears to be a valuable way to study mediation in outcome studies involving small samples.