This doctoral dissertation proposes a Tabu search based metaheuristic optimization method for planning rural MV power distribution networks as well as suburban without preset loop structure, directly including reliability as variable in an objective function. Regarding the time considered in the planning process, the pseudo-dynamic approach is applied. The basic network elements taken into account in planning, i.e. supply transformer substations, MV feeders and switching devices, are modelled by appropriate technical and economic characteristics. Objective function includes fixed and variable costs, supply interruption costs or energy not supplied costs and has to fulfil voltage drop and equipment capacity constraints. Fixed costs consist of investment and maintenance costs while variable costs include linearized resistive losses costs. The estimation of energy not supplied costs and the profitability of construction costs of tie feeders as potential elements of the future network configuration, which enable backup supply, is performed by comparison of their fixed costs with potential decrease of energy not supplied costs in situations without and with backup supply. In both scenarios energy not supplied costs are calculated with the optimal number, type and allocation of switching devices. A special approximation method for calculating and comparison of energy not supplied costs is developed. Regarding the type of optimization problem structure, the mathematical model for MV distribution network planning falls into the class of combinatorial problems which in term of computational complexity represent NP-hard problems. Therefore a metaheuristic Tabu search optimization method is developed and applied. This method is based on the network simplex algorithm structure extended to include not only variable costs, but fixed ones as well. In the introduction the main issues and features of MV distribution network planning are briefly presented, doctoral thesis topic is explained and the scientific contribution of the dissertation is described. The first three chapters deal with main characteristics of operational, short-term and long-term MV distribution network planning and mathematical models and optimization methods for long-term MV distribution network planning. Typical configurations of urban, suburban and rural MV distribution networks are described. Technical and economic characteristics of supply substations and MV feeders as well as physical and mathematical features of the MV distribution network planning models are presented. Objective function structure is analyzed and computational complexity of the optimization algorithms is considered. Classification criteria of the MV distribution network planning models and methods as well as an overview of the published scientific papers are given. Chapter 4 introduces general terms of the reliability theory as well as distribution network reliability indices focusing on reliability worth i.e. energy not supplied costs assessment. Including reliability in a network planning model, both as a preset objective function constraint or explicitly as a variable, is considered. A fast approximation method based on uniformly loaded feeder for determining optimal number and locations for disconnectors or load-breaking disconnectors on the MV feeder for cases without and with a backup supply is developed and presented in Chapter 5. Characteristic outage times are defined. The MV feeder with laterals is substituted by an equivalent one without laterals. It is presumed that backup supply is placed at the end of the feeder. The minimum value of the objective function consisting of energy not supplied costs and switching devices costs is calculated. The difference between the optimal costs, consisting of energy not supplied costs and switching devices costs, for cases without and with backup supply is the criterion of the tie feeder construction profitability. The method is explained with one simple MV feeder example. The accuracy of the method is tested and additional corrective factors for its improvement are introduced. The model and the approximation method for determining the optimal number and locations of switching devices on the MV feeder are improved in the next chapter so that simultaneous installation of disconnectors, load-breaking disconnectors and circuit breakers can be considered and their optimal number and locations can be determined for cases without and with a backup supply. In Chapter 7 the MV feeder with laterals is not substituted with an equivalent feeder without laterals but with an approximate uniformly loaded structure still containing laterals. Thus the installation of switching devices at the beginning of the laterals is enabled. With this improvement the model can be applied in MV network planning because it makes it possible to consider a backup supply in any node of the main feeder, not only the end node. In this way the difference between the optimal costs, consisting of energy not supplied and switching devices costs, in cases without and with a backup supply in any node of the feeder can be calculated. A mathematical model for an optimal MV distribution network routing which includes reliability as variable in the objective function is presented and its possible solution methods are indicated. The network simplex algorithm, structure of which is used as the base for the upgrade of the model by application of Tabu search method, is described in Chapter 9. Firstly, the general simplex method is presented including a version with upper-bound constraints. Then the features of the network simplex algorithm are given in detail and illustrated with simple examples. Chapter 10 deals with the main characteristics of Tabu search method. The simple Tabu search method, as well as its upgrades, one including diversification and intensification strategies and the other including strategic oscillation, are described. The optimal rural MV distribution network routing using Tabu search method is described in Chapter 11. All procedures of the optimization algorithm are presented in detail by relevant pseudocodes and explained with simple examples. An overall flow diagram is given in the appendix. The application of the developed computer code is illustrated with an example of the 10 kV rural distribution network consisting of one supply substation, 30 nodes and 61 potential MV branches. The algorithm run time on the PC with an 1.8 GHz and 1 GB RAM processor was practically instantaneous.