Introduction: Digital interventions offer scalable and accessible options for addressing substance use disorders (SUDs). PocketSkills is a self-guided, 12-week internet-based dialectical behaviour therapy (iDBT) program developed to support individuals seeking treatment for SUDs by teaching DBT skills such as mindfulness, distress tolerance, emotion regulation, and interpersonal effectiveness. While previous research suggests that higher treatment credibility and expectancy predict better outcomes in traditional psychotherapy, their role in digital interventions remains unclear. This study aimed to assess whether baseline treatment credibility and expectancy were associated with improvements in clinical outcomes following iDBT. Methods: A total of 72 participants were initially randomized to one of two treatment conditions: (1) immediate start or (2) a four-week waitlist. The current study includes participants from both conditions but only those who completed both the baseline and 12-week follow-up assessments while engaging with PocketSkills (N = 59). Treatment expectancy and credibility were assessed using the Credibility/Expectancy Questionnaire (CEQ). Outcome variables included depression symptoms (PHQ-9), emotion dysregulation (DERS-SF), distress tolerance (DTS), and mindful awareness (MAAS). Multiple linear regressions were conducted to examine whether credibility and expectancy scores predicted change scores on each outcome measure. Results: Regression results indicated that CEQ scores did not reliably predict changes across most outcomes. The model for depression approached significance (F(2,56) = 2.56, p = .086), with marginal associations for both credibility (β = .281, p = .059) and expectancy (β = –.289, p = .053). The model for emotion dysregulation (DERS-SF) was significant (F(2,56) = 3.68, p = .031), with credibility positively associated with improvement (β = .390, p = .009); expectancy was non-significant (p = .192). No significant results were observed for distress tolerance (F(2,55) = .232, p = .794) or mindfulness (F(2,55) = 1.43, p = .248). Conclusions: Findings suggest that baseline credibility and expectancy may not consistently predict outcomes in digital DBT for SUD. While credibility showed a modest association with improvements in emotion dysregulation, neither credibility or expectancy predicted change in depression, distress tolerance, or mindfulness. These results may be encouraging for emerging e-health interventions, indicating that users can benefit from digital treatments regardless of initial expectations. Further research should explore additional user-level and treatment-specific factors that contribute to digital intervention success.