The principal goal of this research is to develop the optimization procedure for noise barriers, taking into account the efficiency of noise reduction on the protected area, as well as simple, modular design that allows easy manufacturing and overall cost reduction. The optimization is based on numerical procedure that combines a numerical method for calculating the acoustical properties of the barriers and the genetic algorithm used for optimization of barrier design. The weighting function used for evaluation of the results depends on the improvement of acoustical parameters, the manufacturability coefficient and the economic feasibility coefficient. Acoustical properties of the barriers are evaluated according to mean value of sound pressure level reduction, defined as the arithmetic mean of noise level reduction in octave bands of interest (from 100 to 4000 Hz). The manufacturability of a barrier is evaluated with a numerical coefficient that expresses the ratios of the required number of technological steps, the quantities of raw materials used, the time required for use of specific machinery and tools and the overall quantity of barriers that can be manufactured in a unit of time with equal resources. The economic feasibility coefficient reflects the total cost of barrier implementation on a given location. The genetic algorithm used for barrier optimization searches for the best possible construction of the modules that are to be used for assembling the barriers on a given location. The algorithm performs the search in the database of available materials and technological processes that can be applied to certain combination of materials, as well as in the feasibility data. Besides the optimization of module construction, the algorithm also searches for the best possible way of assembling a barrier on a given location, with a predefined number of different elementary modules.