Ad hoc networks are infrastructure-less and self-organizing networks that consist of static or mobile nodes with limited bandwidth, computing ability and energy. These networks are deployed for a wide range of civilian and military applications. Having an efficient and reliable routing protocol for communication between the nodes can be critical. Our goal in this thesis is to exploit Multi-Beam directional Antennas (MBAs) to fisignificantly reduce the end-to-end (E2E) delay in multi-hop ad hoc networks that service multiple traffic flows. We conduct this work in four major steps. First, we explore the benefits of directional antennas, from the standpoint of single-beam ones, for traditional routing protocols. We propose a single-beam directional antenna MAC protocol in the process. Secondly, using Flying Ad hoc Networks (FANETs) as an example, we make the case that MBAs are yet to be exploited for E2E delay reduction. To that end, we propose a multi-beam directional antenna MAC protocol. Third, as a consequence of the case made in the second step, we propose a Mixed Integer Linear Programming (MILP) model that exploits MBAs' capabilities for delay minimization. Solving this model shows that the routes that are selected for the different flows need to have certain key characteristics that depart from the widespread traditional shortest-route philosophy. Based on these characteristics, we design, in the fourth step, an MBA-Delay-Reducing Routing protocol (MBA-DRR) that fully exploits the benefits of MBAs for delay reduction. The benefits of this protocol apply to all types of multi-hop MBA-based ad hoc networks, both mobile and static. As a matter of fact, the evaluation on a multi-flow static scenario shows that MBA-DRR, with a delay of just 4.4 ms, gets very close to the optimal solution that has a delay of 2.5 ms. Comparatively, Reactive-Geographic hybrid Routing (RGR), a shortest-route-based protocol, has a delay of 48 ms. An evaluation on a representative multi-flow mobile scenario shows that, while single-beam directional MAC reduces the E2E latency from 700 ms to 40 ms, and multi-beam directional MAC halves this to 20 ms, our proposed routing protocol further cuts it to 9 ms.