There is little evidence relating to the effects of adding guidance to internet-based gambling interventions. The primary aim was to compare the effectiveness of an online self-directed cognitive-behavioural gambling program (GamblingLess) with and without therapist-delivered guidance. It was hypothesised that, compared to the unguided intervention, the guided intervention would result in superior improvements in gambling symptom severity, urges, frequency, expenditure, psychological distress, quality of life and help-seeking. A two-arm, parallel-group, randomised trial with pragmatic features and three post-baseline evaluations (8 weeks, 12 weeks, 24 months) was conducted with 206 gamblers (106 unguided; 101 guided). Participants in both conditions reported significant improvements in gambling symptom severity, urges, frequency, expenditure, and psychological distress across the evaluation period, even after using intention-to-treat analyses and controlling for other low- and high-intensity help-seeking, as well as clinically significant changes in gambling symptom severity (69% recovered/improved). The guided intervention resulted in additional improvements to urges and frequency, within-group change in quality of life, and somewhat higher rates of clinically significant change (77% cf. 61%). These findings, which support the delivery of this intervention, suggest that guidance may offer some advantages but further research is required to establish when and for whom human support adds value.