The current study examines the understanding of the DSM-IV-MR-J items to assess pathological gambling among adolescents aged 12 to 15, and explores its accuracy. The DSM-IV-MR-J was first administered in the classroom. Participants were assigned to either an experimental or a control group. Participants in the first group were asked to explain the meaning of each DSM-IV-MR-J item during an individual interview. If the item was not properly understood, the investigator corrected the participant's understanding of the item. The questionnaire was then administered a second time. The control group was only submitted to a test-retest procedure. The results showed that 22% of the items were misunderstood. Changes in diagnostic categories emerged on the second administration for both groups. A 20% and 29.4% decrease in the number of problem/pathological gamblers was observed in the experimental and control group. The implications of these results are discussed in terms of the reliability of the DSM-IV-MR-J as a measure of problem gambling among adolescents.