How accurately can the electrically evoked taste threshold be determined? Why have standards of normal values not yet been generally accepted? In order to answer these questions, 80 normal subjects were examined by means of the method of adjustment. The results showed a greater variability of threshold values and side differences than found by other authors. In explanation, it is suggested that the method of adjustment minimizes the biasing effect of the examiner, which normally leads to an underestimation of the threshold variability as well as of the side differences. The method of adjustment is therefore suitable for certain experimental purposes. For clinical use, a simple procedure like Krarups should be preferred. Due to its limited validity, however, only side differences of 100% or more should be considered pathological. A log-scale with rather large units (25% increment steps) is recommended.