Assigning homework is a key component of cognitive behavioural therapy (CBT) and engagement of homework has been shown to predict CBT outcomes for a range of disorders. However, no previous study has empirically examined directly the effect of homework engagement on CBT outcome for problem gambling (PG). To address this gap in the literature this study examined the role of homework along with two non-specific variables (treatment credibility and expectancy), and their prediction of short-term outcome in the context of exposure-based CBT for a group of treatment-seeking problem gamblers. Seventy-four treatment-seekers were invited to participate in a graded cue exposure with response prevention programme developed for problem gambling. The primary outcome measure was the Victorian Gambling Screen Harm to Self sub-scale. Homework engagement data were examined for participants who engaged in the programme and provided outcome data (n = 45). Homework engagement but not treatment credibility or expectancy predicted outcome at post-treatment and at one-month follow-up. The present findings suggest that engaging in homework has a strong association with immediate and short-term outcome, and that improvement of homework engagement has the potential to improve clinical outcomes in CBT targeting PG.