This thesis deals with accurate modelling of energy storage systems in standalone wind energy systems when the reliability and cost optimization are of interest. In recent years, the use of renewable energy resources such as wind and solar has faced a rapid growth due to the environment-friendly nature of these sources on one hand and increasing cost of fosile fuels on the other. In vast countries, there are some remote and unreachable areas which are far from the power system grid. Using renewable energy sources to supply electricity in such areas could have many technical and economical advantages. For example, the cost of expensive transmission lines can be saved. Many renewable energy resources are atmosphere dependent and, thus, have intermittent generation nature. Therefore, in most cases they alone cannot provide an uninterrupted and reliable source of energy. The solution is to use these resources besides other types of energy supplies which are readily dispatchable. Energy storage systems are good candidate for such applications. They can store energy when it exceeds the demand, and release that energy when the generation does not suffice the load. Such hybrid energy systems must be designed in such a way that they have a reasonable reliability at justifiable cost. The reliability of small standalone networks is evaluated by determinig the ability of supplying the demand. In this way, the system adequacy is examined by probablistic techniques based on analytic methods. The disadvantage of analytical methods in evaluating wind systems is the fact that they do not consider the time variation of wind speed which has a significant effect on reliability of these systems. In addition, the existence of storage systems adds to the complexity of problem. In this thesis, the reliability indices are obtained based on analytical methods and time-series simulation of system behavior. In this method, a time series is used to express the energy and power associated with each component. In performing such analysis, the accurate modelling of system components is very important. Specifically, it is shown in this thesis that how the accurate modelling of energy storage systems can affect the reliability indices. A reliability index called loss of load expectation, or LOLE, is obtained for different scenarios and the resultas are compared. Generally, the initial cost of energy systems with renewable resources is more than the cost of traditional systems such as motor-genarator sets. What makes renewable systems more attractive are small running costs and the incentives given by the government. In this regard, the system configutation in terms of the type and size of different resources plays an important role in designing an economically viable system. For example, a wind-only system will be too expensive to supply the load with reasonable reliability. A diesel generator has much lower initial cost, but the penalties associated with pollution and the cost of supplying fuel to it may impose high running cost. Energy storage systems also have relatively high initial cost. Therefore, a comprehensive study is required in order to determine the optimum size of different components in such systems. An optimazation problem is formulated in this thesis based on various constraints and consequently is solved using genetic algorithms. Again, the results for different scenarios are presented and compared with. Keywords: Renewable sources, Energy storage systems, Reliability, Optimum size selection.