To determine whether tinnitus has a masking effect similar to that of general external sounds, the hearing threshold shift when the tinnitus was changed by intravenous administration of lidocaine was measured by contingent negative variation (CNV) audiometry. In a preliminary study, the reliability of CNV audiometry was evaluated in 12 patients with tinnitus. The reproducibility of CNV audiometry was confirmed, based on repeated measurements of CNV audiometry at the tinnitus frequency. The threshold difference between CNV audiometry and pure tone audiometry was -0.76 +/- 1.35 dB, and a high statistical correlation between the threshold of CNV audiometry and pure tone audiometry was evident. The CNV threshold shift at tinnitus frequency after lidocaine injection was measured in 89 ears. In the tinnitus-decreased group, the CNV threshold decreased (4-6 dB) in 22% of cases; however, in the tinnitus-unchanged group, the CNV threshold was little changed. These findings suggest that tinnitus may have a masking effect as is the case with external sounds.