Objective: The study explores agreement on diagnoses and diagnostic criteria for substance use disorders between two structured assessment interviews, the Structured Clinical Interview for the DSM-IV (SCID) and the Alkohol/Drog Diagnos InStrument (ADDIS). Both interviews are compared with a golden standard (GS), based on a LEAD model (Longitudinal, Expert, All Data). Method: Patients were interviewed concerning substance use problems by trained interviewers using SCID and ADDIS separately and blind to each other's results. SCID and ADDIS interviews were compared with each other, and both were compared with a GS. Results: Satisfactory agreement exists between SCID and ADDIS on criteria as well as final diagnostic suggestions, although ADDIS tended to propose dependence diagnoses somewhat more often than SCID. Agreement between SCID and GS is moderate. Sensitivity of SCID is satisfactory, as is specificity for lifetime diagnoses, while specificity for current diagnoses is perfect. ADDIS demonstrates substantial to perfect agreement with GS on dependence diagnoses and moderate agreement on abuse diagnoses (both lifetime and current), as well as showing excellent to perfect overall sensitivity and specificity. Both instruments are in almost perfect agreement with the GS on severity ratings. Conclusion: Both ADDIS and SCID can be used to ensure good standards in the diagnostic assessment of substance use disorders (both alcohol and drugs), with and without psychiatric comorbidity. Significant outcomes. Both SCID and ADDIS are in good agreement with the GS based on a LEAD model concerning substance use disorders.