Much research has been conducted on Mobile Ad Hoc Networks (MANETs) over the past decade due to the rapid advances in wireless communication technology and the wide potential applications of MANETs in many walks of life. A lot of research effort has been devoted to developing efficient MANET routing protocols that exhibit low overhead during path discovery and maintenance. In particular, grid-based routing protocols have been shown to incur a low overhead by exploiting a logical two-dimensional grid structure over the topological terrain which helps identifying paths between source and destination nodes in a cost-effective manner. Most previous grid routing protocols are based on the well-known GRID protocol. This protocol is in turn based on the well-known reactive AODV protocol over a backbone of cell-heads; one cell head from cach cell in the 2D-grid structure. A cell-head is elected in each cell using a specific cell-head election mechanism. Moreover, a data packet travels from source to destination by hopping repeatedly from one cell to another adjacent cell until it reaches its destination. Existing grid-based routing protocols suffer from drawbacks including inefficient election mechanisms that use a high number of control packets causing high end-to-end delay; inefficient use of the transmission range of the nodes by restricting the packet movement to adjacent cells: and the lack of utilizing the available alternative paths. In the first part of this thesis, I propose and evaluate a new cell-head election mechanism in order to reduce overhead due to control packets and improve delivery ratio and end-to-end packet delays, The routing protocol resulting from the new election mechanism is referred to as Efficient Grid based Routing Protocol (EGRP). Our approach is to use regular control packets (including route request and route reply packets) for piggybacking information required for cell-head election as well as a special table at each node for storing minimal information about neighbouring nodes. In existing grid-based protocols, including our proposed EGRP, a packet travels from a cell to another with a restriction that the two cells have to be adjacent. I show that this restriction could affect negatively system performance, especially in sparse environments where there are many empty cells. To address this shortcoming, I suggest an enhancement to EGRP where such a restriction is relaxed. Our simulation results show that this helps reducing overhead and improves packet delivery ratios. In the second part of this thesis, I propose a novel grid-based routing approach which combines the desirable features of both proactive and reactive routing protocols in order to achieve higher performance levels. The approach divides the routing protocol model into two layers. The first is a proactive layer where a cell-based shortest-path tree is constructed. The second is a reactive layer where paths to destinations are established making use of the constructed shortest-path tree. I propose a new protocol called Trec-based Grid Routing (TGRP) based on the proposed two-layer approach. I have developed a mathematical model for studying the performance of TGRP compared to other grid-based routing protocols (GRID and EGRP). The results show the superiority of TGRP especially in terms of reduced overhead and end-to-end packet delays. During the course of my research, I have designed and implemented simulation models and conducted extensive simulation experiments using the NS2 network simulation tool. In these simulations I have analysed the performance of my proposed routing protocols under a variety of network density, traffic load and node mobility conditions. I have collected statistics on the following performance measures: end-to-end packet delay, control overhead, and packet delivery ratio. In most considered scenarios, results reveal that my proposed protocols lead to great improvements in the performance of grid-based protocols with respect to the adopted measures