Power flow in earth return distribution systems typically depends on geographical location and specific earth properties. The planning of such systems has to take into account different operational and safety constraints from conventional distribution systems. This work presents the mathematical modeling and planning of Single Wire Earth Return (SWER) power distribution networks. The SWER load flow is modeled and formulated as an optimization problem. Then by using a heuristic iterative procedure, a planning algorithm is developed for the SWER system. The developed procedure includes optimal feeder routing and overhead conductor selection for both primary and lateral feeders with load growth over several time periods. A 30 node test network extracted from a rural area in Uganda is used to test the algorithm's practical application to give reasonable and consistent results. The model presented can be used in planning SWER networks for areas which have previously not been electrified as well as determining suitable upgrades for existing SWER distribution feeder. The algorithm's mathematical modeling and simulations were done using the General Algebraic Modeling System (GAMS).