A procedure, suitable even for hand calculation, is presented for optimization of single-source water supply systems. The procedure assumes that all nodal demands are known and that minimal hydraulic gradient level requirements at the nodes are specified. The optimization problem is solved through an iterative procedure in which the nodal hydraulic gradient levels are initially assumed and successively corrected to satisfy the cost-head-loss ratio optimality criteria developed in the paper. The diameter is treated as a continuous dependent variable and the odd diameter links are replaced by equivalent commercially available pipes. The procedure is illustrated by a design example. The procedure is applicable to branched, looping, as well as combined networks, and is found to compare favorably with computer-oriented design procedures.