It is important to predict the future available capacity and the remaining useful life (RUL) of battery accurately for the stable driving of electric vehicles in long terms. Therefore, a predicting technique integrating ARMIA and PF are presented in the paper. The ARIMA learns the past ten available battery capacities, which is used to predict the next available capacity. Particle Filter (PF) builds a model based on the past long-term available capacities, which is used to predict the decline law of battery capacity and discover the RUL of battery. The method proposed in this paper is tested by several batteries of NASA. The mean absolute errors of battery5, battery6 and battery7 are 0.006871 Ah, 0.011197631 Ah, and 0.005769204 Ah, respectively. The RUL error of battery B5, B6 and B7 by PF algorithm are 9 cycles, 2 cycles and 6 cycles, respectively.