The advent of colorless and non-directional reconfigurable optical-add-drop multiplexers (ROADMs) will enable flexible pre-deployment of optoelectronic regenerators in future optical networks. Compared to the current practice of installing regenerators only when a circuit needs them, pre-deployment will allow service providers to achieve rapid provisioning and restoration. The pre-deployed regenerators should be concentrated in a selected subset of ROADM sites in order to attain high utilization and to reduce operational costs. We prove that the resulting optimization problem is NP-hard and present an efficient heuristic for this problem that takes into account both the cost of individual circuits (regenerator cost and transmission line system cost) and the probability of a given circuit request, as well as the number of regenerator sites. We provide various methods to reduce the number of regenerator sites, if low probability demands are allowed to have slightly costlier paths. Specific network examples show that the proposed heuristic has near optimal performance under most studied scenarios. We present results for several different cost models. We have also evaluated the heuristic for survivable optical networks, in which a second, disjoint path must be supported for each circuit. An extended version of this paper containing proofs, pseudo-codes and additional experimental results is available online .