To maintain a degrading system, many groups focusing on different tasks may be involved. It is often expected to allocate limited resources (e.g., the budget for technical training) among those maintenance groups to achieve the highest system performance. However, to allocate the resources and implement the maintenance tasks, the sequence of maintenance activities constrained by the system structure has to be considered. This paper provides a model for the maintenance of degrading systems under installation constraints. Based on this model, a nonlinear programming (NLP) model is presented in optimizing the allocation of maintenance resources. In particular, the failure replacement (FR) policy and opportunistic maintenance (OM) policy are considered for a two-module series system. The associated steady-state availabilities of the system are derived for the two cases. To demonstrate the approach in practical use, a numerical example for the FR maintenance policy is provided, and a significance analysis is conducted to provide more insight into the resource allocation issue.