Background: In Turkey, studies of childhood depression have been limited to those using only scales. 

Methods: This is a two-stage cross-sectional epidemiological study. Three schools in Fatih, Istanbul, Turkey were selected by cluster sampling. The study included 1482 students between the fourth and eighth grades. The Child Depression Inventory (CDI) was used for screening in the first stage. According to test results, 320 children were interviewed using the Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version. Depressive disorders were diagnosed in 62 students, constituting the case group. The control group was matched to the case group on school, grade, age, gender, and socioeconomic status (SES). 

Results: The prevalence rate was 4.2% for some form of depressive disorder, 1.55% for major depressive disorder, 1.75% for dysthymic disorder, 0.26% for double depression, and 0.60% for depressive disorder-not otherwise specified. According to the logistic regression analysis, increasing age, having a working mother and low maternal education were all positively associated with CDI score. Low SES, perception of paternal health and relations with father as negatively, low popularity and behavior subscales of the Piers-Harris self-concept scale, and high trait anxiety were associated with the presence of a depressive disorder. 

Limitations: Parents were not included in the diagnostic interviews. 

Conclusions: Dysthymic disorder was the most common depressive disorder. Low maternal education, low SES, dysfunctional interaction with the father, anxiety, and low self-esteem, which were all associated with depression, may be the target of interventions for prevention and treatment of depression.