ABSTRACT Aims This study evaluated the prevalence and reliability of DSM-IV adopted criteria for 3,4-methylenedioxymethamphetamine (MDMA) abuse and dependence with a purpose to determine whether it is best conceptualized within the category of hallucinogens, amphetamines or its own category. Design Test-re-test study. Participants MDMA users (life-time use >5 times) were recruited in St Louis, Miami and Sydney (n = 593). The median life-time MDMA consumption was 50 pills at the baseline. Measurements The computerized Substance Abuse Module for Club Drug (CD-SAM) was used to assess MDMA abuse and dependence. The Discrepancy Interview Protocol (DIP) was used to determine the reasons for the discrepant responses between the two interviews. Reliability of diagnoses, individual diagnostic criteria and withdrawal symptoms was examined using the kappa coefficient (kappa). Findings For baseline data, 15% and 59% met MDMA abuse and dependence, respectively. Substantial test-re-test reliability of the diagnoses was observed consistently across cities (kappa = 0.69). 'Continued use despite knowledge of physical/psychological problems' (87%) and 'withdrawal' (68%) were the two most prevalent dependence criteria. 'Physically hazardous use' was the most prevalent abuse criterion. Six dependence criteria and all abuse criteria were reported reliably across cities (kappa: 0.53-0.77). Seventeen of 19 withdrawal symptoms showed consistency in the reliability across cities. The most commonly reported reason for discrepant responses was 'interpretation of question changed'. Only a small proportion of the total discrepancies were attributed to lying or social desirability. Conclusion The adopted DSM-IV diagnostic classification for MDMA abuse and dependence was moderately reliable across cities. Findings on MDMA withdrawal support the argument that MDMA should be separated from other hallucinogens in DSM.