Service Providers have come to realize the importance of traffic engineering techniques because they allow them to optimize utilization of network resources and bring more revenues. In fact, one of the biggest reasons for the success of MPLS as a technology is its ability to implement traffic engineering at a low cost, especially in terms of operational expenditure. Dynamic computation of optimal paths through the network, dynamic provisioning of tunnels, and proactive distribution of traffic between the tunnels are examples of how MPLS pushes the intelligence for operation of a traffic engineered network into the network itself and away from proprietary central management stations, thus allowing for the building of efficient and cost-effective multi-vendor networks. In this chapter we will discuss what traffic engineering is, how it applies to GMPLS-controlled transport networks, and how it is different from traffic engineering in packet switching networks. We will define the notions of a transport service, network controller, data switch, and control channel, as well as control interface, data interface, data link, and link attributes in the context of traffic engineering. We will go on to discuss ways that GMPLS nodes learn about other nodes and links, so that path computation elements can determine optimal paths for services with specified attributes. Finally, we will analyze the peculiarities of traffic engineering on networks that span multiple regions and administrative domains.