Lithium-ion batteries have emerged as a mainstream source to store energy in electrified vehicles due to pivotal features of high energy density, long cycle life and low self-discharge. The continuous monitoring of state of charge (SOC) of a lithium-ion battery is essential to avoid over-charging or over-discharging in order to ensure safe operation as well as to reduce its average life cycle cost. However, an accurate SOC estimation of lithiumion battery has become a major challenge in the automotive industry. In this paper, k-nearest neighbours (kNN) concept have been employed to estimate the SOC, based on the measured voltage, current and previous SOC. A Gaussian filter is employed to minimize the errors in the kNN-based SOC estimation. The effectiveness of the proposed hybrid method is verified on the experimental data of the lithium-ion battery under different standard driving schedules and temperatures. Results show that the proposed hybrid approach outperforms other conventional SOC approaches with better accuracy under federal urban and highway driving schedules.