OBJECTIVE: To assess the effect of food insecurity on perinatal depression in rural Ethiopia. DESIGN: We used a prospective cohort in which food insecurity was considered as primary exposure and perinatal depression as an outcome. Food insecurity at baseline (in the period of 8-24 weeks of pregnancy) was measured using the Household Food Insecurity Access Scale (HFIAS), and perinatal depression at follow-up (in 32-36 weeks of pregnancy) was measured using a Patient Health Questionnaire (PHQ-9). We used multivariable regression to assess the effect of food insecurity on the prevalence of perinatal depression. We explored food insecurity's direct and indirect impacts on perinatal depression using structural equation modelling (SEM). SETTING: This paper used data from the Butajira Nutrition, Mental Health and Pregnancy (BUNMAP) cohort established under the Butajira Health and Demographic Surveillance Site (BHDSS). PARTICIPANTS: Seven hundred and fifty-five pregnant women. RESULTS: Among the study participants, 50 % were food-insecure, and about one-third were depressed at 32-36 follow-up. In SEM, higher values of baseline food insecurity, depressive symptoms and state-trait anxiety (STA) were positively and significantly associated with perinatal depression. The direct impact of food insecurity on perinatal depression accounts for 42 % of the total effect, and the rest accounted for the indirect effect through baseline depression (42 %) and STA (16 %). CONCLUSION: The significant effect of food insecurity at baseline on perinatal depression and the indirect effect of baseline food insecurity through baseline anxiety and depression in the current study implies the importance of tailored interventions for pregnant women that consider food insecurity and psychosocial problems.