Aims Evidence from randomised trials analysing effects of depression treatment on glycaemic control in group comparisons is inconsistent. The aim of this study was to test if the reduction of depressive symptoms would explain improved glycaemic control irrespective of treatment groups. Methods The DIAMOS study tested effects of cognitive-behavioural therapy (CBT) versus usual care on depressive symptoms in a 12-month prospective trial; HbA 1c was a secondary outcome. Since the results suggested superiority of CBT for improving depressive symptoms, but not HbA 1c , we conducted this secondary analysis to test if reduction of depressive symptoms could explain improved glycaemic control when assessed irrespective of treatment group affiliation. Reduction of depressive symptoms was assessed using baseline-to-follow-up changes in the Center for Epidemiologic Studies Depression Scale (CES-D). We used multiple regression analyses, adjusting for baseline HbA 1c and depression, group affiliation and covariates, to assess associations between reduction of depressive symptoms and follow-up HbA 1c . Results 181 participants provided eligible data. Depressive symptoms decreased between baseline and follow-up by averagely −5.1±11.8 CES-D points. Greater reduction of depressive symptoms predicted greater improvement of HbA 1c at follow-up, while adjusting for baseline HbA 1c and covariates (Beta=−0.24, P =0.004). Additionally, patients with greater reduction of depressive symptoms were more likely to reach in-target HbA 1c ( P =0.023). Conclusions The findings suggest that reduction of depressive symptoms can explain improved glycaemic control. Behavioural treatments might aim to improve both affective and glycaemic outcomes.