BACKGROUND: clinicians have long noticed a correlation between physiological markers of inflammation and depression. The best-known example is the activation of the hypothalamus-pituitary-adrenal axis and cortisol secretion; however more recent studies have demonstrated increased salivary prostaglandins and plasma acute phase proteins in depressed patients. To date four randomised controlled trials have used celecoxib or rofecoxib as adjuncts to serotonin selective reuptake inhibitors in the treatment of depression. These suggested a statistically significant decrease in depressive symptoms in the patients taking NSAIDs and SSRIs, compared to patients taking SSRIs alone. Interpretation of these results is limited by the small sample size and short duration of these preliminary studies. The research only considers depressed patients receiving treatment in secondary care; no study has examined the effectiveness of NSAIDs as an adjunct in primary care, even though most cases of depression in the UK are managed in the community by general practitioners. PROPOSAL: we propose a multi-centre double-blinded randomised controlled trial with two objectives: to determine whether citalopram plus celecoxib dual therapy achieves a greater reduction in depressive symptoms (quantified using the Hamilton Depression Rating Scale (HDRS)) within four weeks, compared to citalopram monotherapy; and to determine whether citalopram plus celecoxib dual therapy is more likely to achieve remission (HDRS score ?7) of moderate to severe depression within six months, compared with citalopram monotherapy. The endpoints will be the reduction in HDRS score after 4 weeks of treatment, and the HDRS score after 26 weeks of treatment. The study will enrol 452 participants from general practices who have a moderate or severe, current or recurrent major depressive episode when medication with an SSRI is considered. The study population will be stratified according to age, sex, HDRS score, age of onset of first episode, number of previous depressive episodes and duration of current episode. The population will then be randomised into two groups. Subjects will be interviewed to determine HDRS score, measure blood pressure, count pills and discuss side-effects. This will occur weekly for the first four weeks, and every four weeks thereafter.