This paper presents a series of 12 cases of chronic tinnitus patients who participated in 4 weeks of auditory discrimination training either close to or far removed from the tinnitus frequency. The training was based on the assumption that tinnitus is related to a shift of the representation of the tinnitus frequency in auditory cortex outside of the normal tonotopic map and that training close to but not removed from the tinnitus frequency should result in a reduction in the severity of the tinnitus. Tinnitus severity was measured 4 times per day during the entire treatment and other tinnitus-related variables were assessed 1 week before and 1 month posttreatment. The comparison of the training close to as compared to remote from the tinnitus frequency did not yield a statistically significant difference. However, a post hoc analysis revealed that patients who engaged in regular training as compared to those who practiced irregularly were significantly more successful in reducing tinnitus severity independent of the trained frequencies. Treatment success was best predicted by days of training and general activity levels. The data suggest that auditory discrimination training shows a dose response effect irrespective of training location and that treatment success is also related to psychological variables. For more substantial changes in multiple variables an extended training period with additional consideration of emotional variables would be necessary. In addition, controls for nonspecific training effects need to be implemented.