Distribution system reliability has been improved economically reducing energy loss by performing DG penetration and capacitor placement. Two novel evolutionary computation techniques called Fish Electrolocation Optimization (FEO) and Modified Cultural Algorithm (MCA) have been implemented for performing cost based reliability improvement in radial distribution system. Reliability improvement, energy loss and total cost minimization have been studied as single objective optimization by real coded Genetic Algorithm (rcGA), Particle Swarm Optimization (PSO), Fish Electrolocation Optimization (FEO) and Modified Cultural Algorithm (MCA). The results obtained by FEO and MCA are promising. Multi-objective optimization has been performed considering distribution system reliability index SAIFI, SAIDI and AENS as well as total cost and energy loss. Pareto optimal solutions have been found out by effective MCA and FEO. It has been observed that the non dominated solutions obtained by FEO are better than the solutions achieved by MCA.