1. Introduction In result of recent human activities, concentration of greenhouse gases has been increased and consequently temperature of the Earth’, s surface has been raised. Also, it is expected that temperature of the Earth’, s surface will be increased in the future (IPCC, 2001,IPCC, 2013). This increase will also influence other climate variables such as precipitation. For active adaptation strategy, it is necessary to assess the potential future climate change impacts. To obtain future climate scenarios, the most common tools are GCMs. since resolution of the GCM outputs is course, it is necessary to downscale its outputs. Among downscaling techniques, Weather Generator (WG) method have several unique advantages, such as: 1-Changes in various statistics, predicted by the GCMs experiments, may be preserved in downscaled series (Khazaei et al., 2013,Semenov et al., 1998). 2-WGs produce long term series that decrease uncertainty of climate variability (Chapman, 1998,Semenov et al., 1998). ARMA model and daily LARS-WG model are two stochastic WGs. LARSWG frequently used for climate change impact assessment. But despite the specific capabilities of the ARMA model to assess the impacts of climate change on an annual scale, this model has been rarely considered in previous climate change studies. The reason is that it is not clear how skewed series can be downscaled using this model (Khazaei et al., 2013). Precipitation series on the daily and monthly time-scales is often skewed, but annual rainfall in many areas has a normal distribution. In this paper, the performance of the annual ARMA model and the LARS-WG daily model for generating annual rainfall and temperature series is compared. Then, the ARMA model is utilized to downscale GCM outputs and climate change impact assessment...