Under suitable conditions, artificial in situ aeration is an appropriate and economical method of alleviating water pollution. Regional planning for the control of water pollution can properly consider the use of artificial aeration on a river network in conjunction with other pollution abatement techniques. Application of optimization methods can yield significant savings in the aeration energy consumption required to produce a given level of impact on a polluted river network. The confluence of two rivers, each receiving effluent discharges and subject to artificial aeration control, is first investigated. It is shown that the optimal feedback aeration strategy in one branch of the river system not only depends on measurements of the water quality in that branch, but also on the water quality measurements in the second branch. With these results as the foundation, the optimal aeration control strategy is then determined for a generalized river network modeled as a graphical tree. Various properties of the optimal aeration policy are illustrated by computational examples.