Maintenance planning for deteriorating transportation infrastructures is quite challenging because multiple maintenance actions with complex effects are usually required. This paper proposes a novel analytical model for deteriorating infrastructures with consideration of multiple types of preventive maintenance actions which have complex effects. Upon each inspection, a decision maker needs to decide whether preventive maintenance is needed, and what type of preventive maintenance action is appropriate if it is desirable. We formulate the maintenance optimization problem as a finite-horizon Markov decision process and investigate the structural properties of the optimal maintenance policies by minimizing the expected cost-To-go function. Computational study is provided to demonstrate the monotonically non-decreasing property of the optimal policies in condition and age using real-world pavement deterioration data. The monotone structure of the optimal policies is appealing in practice, because it can save the computational efforts and facilitates implementation.