Background: Antidepressants are used frequently in non-fatal self-poisoning. There are national guidelines for prescribing antidepressants. There have been few investigations of how non-fatal self-poisoning with antidepressants varies in relation to prescribing and to patient characteristics. Methods: A comparative study of the use of specific antidepressants (amitriptyline and dosulepin (tricyclics), citalopram, fluoxetine, paroxetine and sertraline (selective serotonin reuptake inhibitors) and venlafaxine (serotonin norepinephrine reuptake inhibitor)) for non-fatal self-poisoning (episode-based), relative to prescribing, in three centres in England, 2004 to 2006. Results: There was marked variation between centres in the ratio of rates of self-poisoning to prescribing for specific antidepressants. Higher rates of self-poisoning relative to prescribing for all antidepressants combined, and for venlafaxine, were found in the centre with greater proportions of patients with a history of self-harm and/or previous psychiatric treatment. Within each centre, higher rates of self-poisoning relative to prescribing were found for citalopram and fluoxetine than amitriptyline. However, rates of self-poisoning relative to prescribing for either amitriptyline or dosulepin were also similar to sertraline, which is of concern given the known toxicity of tricyclics. Limitations: An ecological study, where prescriptions were for all indications and not specifically for the patients who self-poisoned. Conclusions: Marked differences found in ratios of self-poisoning with antidepressants to levels of prescribing, in three centres in England, are likely to reflect differences in both prescribing practices (despite clear national guidance) and patient characteristics. Risk of overdose and toxicity should be considered when local prescribing policy and clinical practice relating to antidepressants are under review. © 2009 Elsevier B.V. All rights reserved.