Electroneurogram (ENG) recording techniques benefit from the use of tripolar cuffs because they assist in reducing interference from sources outside the cuff. However, in practice the performance of ENG amplifier configurations, such as the quasi-tripole and the true-tripole, has been widely reported to be degraded due to the departure of the tripolar cuff from ideal behavior. This paper establishes the presence of cuff imbalance and investigates its relationship to cuff asymmetry, cuff end-effects and interference source proximity. The paper also presents a comparison of the aforementioned amplifier configurations with a new alternative, termed the adaptive-tripole, developed to automatically compensate for cuff imbalance. The output signal-to-interference ratio of the three amplifier configurations were compared in vivo for two interference signals (stimulus artifact and M-wave) superimposed on compound action potentials. The experiments showed (for the first time) that the two interference signals result in different cuff imbalance values. Nevertheless, even with two distinct cuff imbalances present, the adaptive-tripole performed better than the other two systems in 61.9% of the trials.