Background. Although the burden of emotional disorders is very high, mental health care is only available to a minority of patients. The literature suggests that self-help strategies, both bibliotherapy and self-help groups alike, are effective for various, less serious complaints but it is unclear whether available data support a role for self-help in treatment protocols for patients with clinically significant emotional disorders.

Method. We searched the literature with a focus on ‘anxiety’ and/or ‘depressive disorder’. Standardized assessment of diagnosis or symptoms and randomized controlled trials were inclusion criteria for a meta-analysis.

Results. The mean effect size of self-help (mainly bibliotherapy) v. control conditions is 0·84, and 0·76 for follow-up; the effect sizes of self-help v. treatment are −0·03 and −0·07 respectively. A longer treatment period is more effective.

Conclusions. Bibliotherapy for clinically significant emotional disorders is more effective than waiting list or no treatment conditions. The dearth of studies on self-help groups for emotional disorders does not permit an evidence-based conclusion concerning the effects of self-help groups. No difference was found between bibliotherapy and psychiatric treatment of relatively short duration.