In a previous paper (Ermakov, S. V.; et al. Anal. Chem. 1994, 66, 4034-4042) it was reported that, when weak acids and bases in a background electrolyte containing a strong co-ion were analyzed at high sample loads (comparable to those of the buffering ion), the analyte peak split into two zones, representing the same species existing in its charged and uncharged forms, The same approach has been here applied to amphoteric species, particularly amino acids, Under similar conditions (high sample load, strong titrant ion) and at operative pH values close to the pi of the analyte, it is shown that the sample can be split into three components, representing the three different charge states of the ampholyte: its cationic, its anionic, and its zwitterionic forms, In the case of His, that the three forms indeed represented the same sample ion was demonstrated by spectral analysis of each peak, This is due to the fact that the strong titrant ion, present in the background electrolyte, can penetrate the sample zone and titrate it not only to the pi value but beyond it, This induces a spatial and temporal pH gradient within the sample zone, increasing from cathode to anode, i.e., having a slope opposite to Rilbe's law of pH monotony, according to which, under focusing conditions, the pH gradient should be a monotonically increasing function from anode to cathode, As a result this phenomenon of peak splitting can be considered, in a broad sense, a ''defocusing'' phenomenon. Computer modeling of this phenomenon showed theoretical profiles in good agreement with experimental ones. Additionally, for peak splitting, the ampholyte should be a a ''good carrier ampholyte'', i.e., exhibit a good buffering capacity and conductivity at the pI, a condition satisfied by a steep incline of the titration curve surrounding the pI value. ''Poor carrier ampholytes'' with wide plateaus in titration curve close to zero charge have a lower probability for three-peak splitting, since they cannot be titrated past their pI value.