Routing is a key aspect of modern computer networks and no other area presents as much challenge to routing as Mobile Ad Hoc Networks or MANETs. These networks are formed on demand and require no centralized control or pre-existing infrastructure. Nodes in MANETs communicate through an unreliable wireless medium and can be mobile. More so, nodes are typically low powered devices with a limited radio range. Due to all these factors, network topology in MANETs can change unpredictably, putting considerable load on the routing protocol. Routing in MANETs has been a hotbed of research activity due to its challenging nature and futuristic applications of such networks. Many routing protocols have been developed over the years and target different operational scenarios with efficiency. Broadly, these schemes are classified as either reactive or proactive. In their pure implementation, neither offers good operational performance and typically a hybrid protocol leveraging advantages of both is often used. In this thesis, a novel hybrid routing algorithm is presented for MANETs. Gradient based Zone Routing or GraZoR combines the advantages of zone based approaches with those of gradient based reactive schemes to achieve robust routing in large scale MANETs. The network consists of 'Zones of Influence' centered on respective nodes. These zones are hop count gradients with estimated cost to reach the zone center directly proportional to a node's distance from it. Routing towards a node in its zone is performed proactively while a reactive approach is used to establish routing knowledge towards a node outside its zone. The algorithm combines the two approaches seamlessly to achieve efficiency and well as robustness. The algorithm is implemented in a MANET simulator and found to achieve effective routing in large MANETS.