AbstractPsychiatric research increasingly emphasizes the diagnosis of symptoms and syndromes on a longitudinal basis. This study tests the reliability of lifetime DSM‐IV diagnoses of alcohol, cannabis, cocaine and opiate dependence. The CIDI‐SAM was administered at intervals not less than six months apart to a multi‐site sample of 201 clinical respondents. The reliability of lifetime diagnosis of the syndromes, of the criteria which constitute the syndromes, and of the ages of onset reported for the criteria and for the dependence syndromes as a whole, were studied and the effects of patient characteristics suspected to degrade reliability were examined. There was generally good agreement, statistically, at both the syndrome and criterion level between the two interviews. Lifetime diagnoses for three of the drugs–alcohol, cannabis and opiates–were made at or near levels of agreement generally considered excellent under less strict testing conditions, and cocaine dependence was only marginally below this level. Most criteria showed good reliability and all delivered about equal results when averaged across the four substances, although a relationship between reliability and centrality of the symptom to the individual drug abuse pattern was found. Age of onset was almost uniformly highly reliable. Most patient characteristics bore no detectable relationship to reliability, although patients with multiple drug use patterns may warrant more careful probing by interviewers. Overall, these data indicate that lifetime symptoms and diagnoses can be queried reliably, although they must be reported with less confidence than current slate