Mental health disorders such as depression and anxiety are leading contributors to the global disease burden. However, a large proportion of people with mental health disorders remain untreated, often due to a lack of nearby services, logistic barriers (e.g., work schedules), stigma, the cost of treatment, and other barriers. Internet-delivered cognitive behavioural therapy (iCBT) without therapist guidance represents a promising solution to this problem, as it is cost-effective, convenient for users, and possible to implement on a large scale. Various persuasive design elements—features or design principles intended to make interventions more engaging—have been proposed and implemented to improve adherence and outcomes among users of unguided iCBT. However, no review to date has determined the extent to which persuasive design elements predict the efficacy of unguided iCBT interventions. The present study consisted of a systematic review, meta-analysis, and meta-regression of unguided iCBT interventions for depression and anxiety. A search of five databases and a hand search yielded a total of 41 articles describing 45 comparisons (N = 10,301) of iCBT conditions and control conditions. The meta-analyses revealed a small to moderate weighted mean effect size (Hedges’ g = 0.29) for unguided iCBT for depression and a moderate effect size (Hedges’ g = 0.48) for unguided iCBT for anxiety. Meta-regression analyses indicated that greater use of persuasive design predicted greater efficacy in unguided iCBT for depression but not anxiety. However, this research was exploratory in nature, and further research will be required to clarify the role of persuasive design in unguided iCBT and its relationship to treatment outcomes.