The study was based on the clinical data summaries ("item sheets") of children who attended the Maudsley Hospital during the late 1960s and early 1970s. These summaries were used to identify a group of 80 child and adolescent psychiatric patients with an operationally defined depressive syndrome. The depressed children were individually matched with 80 non-depressed psychiatric controls on demographic variables and non-depressive childhood symptoms by a computer algorithm. At follow-up, on average 18 years after the initial contact, information was obtained on the adult psychiatric status of 82% of the total sample. Adult assessments were made "blind" to case/control status, and included standardized measures of "lifetime" psychiatric disorder and psychosocial functioning. The depressed group was at increased risk for affective disorder in adult life, and had elevated risks of psychiatric hospitalization and psychiatric treatment. Depressed children were no more likely than control children to have non-depressive adult psychiatric disorders. These findings suggest that there is substantial specificity in the continuity of affective disturbances between childhood and adult life.