Assessing the prevalence of the comorbidity of psychiatric and addictive disease using epidemiologic methods results in artifactually high rates. Use of a clinical sample will yield falsely high rates, because substance use is associated with exacerbation of mental illness. Cross sectional design will inflate rates of psychiatric comorbidity in addicts, who attribute substance use to psychological symptoms until well into recovery. Application of exclusionary criteria for independent diagnosis is subject to investigator bias, particularly about the unproven yet popular "self-medication" hypothesis. The psychiatric symptoms which are common in active addiction generally clear within weeks to months of treatment for addiction but do not respond to standard psychopharmacologic treatment for primary mental illness. When lengthy follow up periods are employed, substance induced psychiatric syndromes typically resolve. We conclude that while patients treated in psychiatric settings often have comorbid and independent addictive illness, patients treated in addiction settings uncommonly have comorbid psychiatric illness despite common psychiatric symptoms.