In this paper, a new three-dimensional path planning approach with obstacle avoidance for UAVs is proposed. The aim is to provide a computationally-fast on-board sub-optimal solution for collision- free path planning in static environments. The optimal 3D path is an NP (non-deterministic polynomial-time) hard problem which may be solved numerically by global optimization algorithms such as the Particle Swarm Optimization (PSO). Application of PSO to the 3D path plan- ning class of problems faces typical challenges such slow convergence rate. It is shown that the performance may be improved markedly by imple- menting a novel parallel approach and incorporation of new termination conditions. Moreover, the exploration and exploitation parameters are optimized to nd a reasonably short, smooth, and safe path connecting the way-points. As an additional precaution to avoid collisions, obstacle dimensions are artificially slightly enlarged. To verify the robustness of the algorithm, several simulations are carried out by varying the num- ber of obstacles, their volume, and location in space. A certain number of simulations exploiting the random nature of PSO are performed to highlight the computational efficiency and the robustness of this new approach.