BACKGROUND: Accurate assessment of suicidality is of major importance. We aimed to evaluate trained clinicians' ability to assess suicidality against a structured assessment made by trained raters. METHOD: Treating clinicians classified 218 adolescent psychiatric outpatients suffering from a depressive mood disorder into three classes: 1-no suicidal ideation, 2-suicidal ideation, no suicidal acts, 3-suicidal or self-harming acts. This classification was compared with a classification with identical content derived from the Kiddie Schedule for Affective Disorders and Schizophrenia (K-SADS-PL) made by trained raters. The convergence was assessed by kappa- and weighted kappa tests. RESULTS: The clinicians' classification to class 1 (no suicidal ideation) was 85%, class 2 (suicidal ideation) 50%, and class 3 (suicidal acts) 10% concurrent with the K-SADS evaluation (gamma2 = 37.1, df 4, p = 0.000). Weighted kappa for the agreement of the measures was 0.335 (CI = 0.198-0.471, p < 0.0001). The clinicians under-detected suicidal and self-harm acts, but over-detected suicidal ideation. CONCLUSION: There was only a modest agreement between the trained clinicians' suicidality evaluation and the K-SADS evaluation, especially concerning suicidal or self-harming acts. We suggest a wider use of structured scales in clinical and research settings to improve reliable detection of adolescents with suicidality.