Background: Recent treatment studies with cognitive behavioural therapy for insomnia (CBT-I) have demonstrated effects on both sleep problems and depression. Two previous studies have indicated that the beneficial effect from CBT-I on depression may come through improved sleep, although insomnia severity during treatment had not previously been investigated as a mediator. Aims: Our aim was to investigate if insomnia severity during treatment mediated between CBT-I and depression severity after treatment, in a sample with co-morbid insomnia and depressive symptomology. We also examined whether depressive severity during treatment mediated between CBT-I and insomnia after treatment. Method: The participants were recruited from advertisements and fulfilled criteria for insomnia diagnosis, and had depressive symptomatology (Beck Depression Inventory-second edition: BDI-II > 13). Two-thirds of the participants were diagnosed with major depressive disorder. The participants received four biweekly group sessions of CBT-I or relaxation training (active control). Insomnia severity (Insomnia Severity Index) and depressive severity (BDI-II) were measured at baseline, mid-treatment, post-treatment and 6-month follow-up. The mid-treatment measures were used as mediators. Results: Mediational analyses demonstrated a significant reciprocal relationship between insomnia severity and depressive severity throughout CBT-I, although mid-treatment insomnia had a stronger effect on depression than mid-treatment depression had on insomnia. The results were similar for both post-treatment and follow-up. Discussion: Some improvement in depressive severity after CBT-I is explained by improved sleep. The findings emphasize the importance of making comorbid insomnia a treatment focus in its own right.