This paper presents an approach for the joint optimization of maintenance and improvements of the components of a network of infrastructure facilities. In the literature, these two problems have often been handled separately, probably because the problems seem quite different. However, these decisions (maintenance and improvement) are not independent due to the presence of trade-offs between the two sets of policies. We develop a Markov decision model for the joint optimization of maintenance and improvement, thus improving the budget allocation among facilities in the network between the two sets of activities and within each set. The model is used to solve for steady-state policies, but relaxes the assumption of age-homogeneous condition-state transition probabilities, which has been criticized in the literature. Moreover, the model allows for the possibility of not exhausting the annual budget available every year, so that part of it can be spent more efficiently in later years. The paper includes a case study that demonstrates that substantial savings can be achieved through the joint optimization of maintenance and improvement policies.