: Today, synthetic aperture radars have wide applications in aerial photography and ground surface monitoring. One of the main challenges in implementation of these radars is signal domain mapping (raw data) to the image domain. Several algorithms have been provided for image formation in synthetic aperture radars that are different in terms of efficiency and computational complexity. These algorithms can be implemented in software and hardware. However, due to the relatively large amount of processing in real-time applications, the hardware implementation must be used.Hardware implementation is executable in floating-point and fixed-point methods. Fixed-point implementation has less hardware resources, more speed and low power consumption. The most important subject on the fixed-point implementation is the selection of an appropriate word length for signals to have an accurate calculation. By calculating and applying the optimal word length for signals and algorithm calculations, we can have an acceptable image quality, reduce hardware resources required and increase processing speed. In this thesis, RDA algorithm which is the most popular and efficient image formation algorithm is used and the optimal word length for its hardware implementation is determined. For this purpose, RDA algorithm is first simulated as floating-point by MATLAB software and by using raw data of ERS-2 satellite radar. Then, by generating the fixed-point model of algorithm functions such as the fast Fourier transform function and using fixed-point signals, the fixed-point model of the algorithm is extracted. In the next step, the fixed-point model is simulated with different word lengths and by comparing its output with the floating-point image by the help of an universal image quality index , the minimum word length required can be obtained to achieve an image quality close to the floating point image. Then, in order to study the implementation effects of fixed-point more accurately, estimate the required hardware resources and working frequency, the hardware model of the system is generated by the use of system generator tools of Xilinx Inc. in the Simulink environment. Finally, the results of the simulation and analysis for implementation on a FPGA of Virtex-5 family are presented. Keywords: Synthetic Aperture Radar, Hardware Implementation, Fixed-Point model, System Generator