Abstract: In battery management systems, there are always colored noises in the sampled battery current signals and voltage signals, which make it hard to achieve the accurate battery state of charge estimation. Regarding these noises as distributions, an H_∞ observer with current debasing for online batter state of charge (SoC) estimation is proposed in this paper. Firstly, the battery stated model with current debasing and noise distribution is built. Secondly, H_∞ observer is designed with current debasing. The estimation accuracy, performance, robust to model errors and parameter adaptation of the observer are analyzed by simulation. At last, experiment results demonstrate its effectiveness.