The unrestricted access to the energy sources is one of the most important elements conditioning fast development of civilisation. However, in view of the climate changes caused by unlimited use of fossil fuels further development is greatly dependent on the effective utilisation of renewable energy sources. One of the most important energy source which utilisation is relatively easy which is not burden significantly the environment is the wind energy. Nowadays, energy production from wind is dominated by huge horizontal axis units (horizontal axis wind turbines – HAWT) which possess relatively high efficiency, however this construction suffers from high starting wind speed and relatively low cut off wind speed. Moreover, this type of wind turbines generates high noise and have significant impact on the environment. Finally, connecting this type of wind turbines into the energy system is very problematic due to random character of energy production. Some alternative to huge HAWTs may be drag-based vertical axis units (Vertical Axis Wind Turbine – VAWT) like the Savonius wind turbine.The Savonius wind turbine offers low efficiency but their availability to produce electric power is higher than traditional HWAT, because they starts at lower wind speeds and can still operate at higher wind speeds comparing to HAWTs. Moreover, this wind turbine is omnidirectional, completely insensitive on the wind turbulence, it produces energy at lower rotational speeds, that is why it can be considered as a power generation units for municipal computations. Paper presents methodology of optimisation of the shape of Savonius wind turbine blades using Bezier splines and Evolutionary Algorithm. This methodology allowed us to increase significantly the rotor performance comparing to the traditional solutions which are considered as an experimentally optimised.