A method for the prediction of remaining useful life (RUL) of li-ion battery for battery management system (BMS) is presented in this thesis work. Estimation of State of charge (SOC), state of health (SOH) and RUL are important tasks in BMS. For the prediction of RUL, information of aging parameter is required. Previously, in RUL prediction methods, aging parameter information was based on the battery tests data. For online estimation of RUL, ageing parameter is required and BMS does have a sensor system which could measure the battery ageing parameter online. So estimation techniques are utilized for the estimation of ageing parameters. Previously different methods were used such as Fuzzy logic, artificial neural network (ANN) and Kalman filters. Fuzzy logic and ANN methods are computationally heavy to implement them in BMS of EVs. Kalman filters methods are not very accurate for the estimation of an ageing parameter because batteries are highly nonlinear system and Kalman filters can only capture nonlinearity up to 3^rd order Taylor series expression. This thesis work presents an algorithm for the prediction of RUL, which is based on the information of SOH. To predict the behaviour of the battery, Thevenin battery model is selected and internal resistance of the battery is considered as an ageing parameter. Internal resistance changes due to change in temperature. Particle filtering framework is implemented for estimation and is found that unscented particle filter gives better estimation of ageing parameter because it adapts the ageing parameter with respect to temperature change. RUL prediction consists of two steps. In first step, SOC and ageing parameter is estimated and then SOH is calculated from ageing parameter. In second step, RUL is predicted from the information of SOH. Variable load and effect of temperature are also considered for the estimation of ageing parameter. SOH is obtained from the ageing parameter and then a degradation model is developed from the information of SOH. Remaining cycles of the battery are then predicted from this degradation model by using particle filter.