Cognitive behavioural therapy self-help is an effective intervention for a range of common mental health difficulties. However the extent to which effectiveness may vary by type of support--guided, minimal contact, self-administered--has not been extensively considered. This review identifies the impact of support on the effectiveness of written cognitive behavioural self-help and further explores the extent to which effectiveness varies across mental health condition by type of support provided. Randomised controlled trials were identified by searching relevant bibliographic databases, clinical trials registers, conference proceedings and expert contact. 38 studies were included in the meta-analysis yielding a statistically significant overall mean effect size (Hedges' g=-0.49). Overall effect size did not significantly differ by type of support (Q=0.85, df=2, p=0.65) (guided: Hedges' g=-0.53; minimal contact: Hedges' g=-0.55; self-administered: Hedges' g=-0.42). For guided and self-administered types of support, planned comparisons revealed a trend for effect size to vary by mental health condition and for guided CBT self-help the modality of support was significant (Q=6.32, df=2, p=0.04), with the largest effect size associated with telephone delivery (Hedges' g=-0.91). Additional moderator analysis was undertaken for depression given the number of available studies. Regardless of higher baseline levels of severity the effect size for minimal contact was greater than for guided support. Greater consideration should be given to the potential that type of support may be related to the effectiveness of written cognitive behavioural self-help and that this may vary across mental health condition. Findings from this systematic review make several recommendations to inform future research.