This paper introduces Epidemic Routing, a protocol for delivering messages in partially-connected ad hoc networks where a connected path from source to destination may not exist. Existing ad hoc routing protocols assume a connected path, but in scenarios with limited connectivity or network partitions, this assumption is invalid. Epidemic Routing ensures eventual message delivery through random pairwise exchanges among mobile hosts, maximizing delivery rate, minimizing latency, and reducing resource consumption. The protocol distributes messages to carriers within connected portions of the network, allowing messages to spread through transitive transmission. It leverages the theory of epidemic algorithms, where random data exchange ensures eventual consistency. Epidemic Routing is implemented in the Monarch simulator, showing that it can deliver 100% of messages with reasonable resource consumption in scenarios where existing protocols fail. The protocol is robust to varying network conditions, including low transmission ranges and limited node mobility. It balances resource consumption and delivery rate by limiting hop count and buffer size. The paper evaluates Epidemic Routing under various scenarios, demonstrating its effectiveness in delivering messages even when traditional routing protocols fail. The protocol is sensitive to node density and transmission coverage, and its performance is influenced by buffer size and message priorities. Future work includes exploring hybrid routing techniques and leveraging positioning information to improve resource utilization. Epidemic Routing is suitable for applications such as mobile sensor networks and disaster recovery, where reliable message delivery is critical despite limited connectivity. The protocol ensures eventual delivery by distributing messages through multiple carriers, making it a viable solution for ad hoc networks with intermittent connectivity.This paper introduces Epidemic Routing, a protocol for delivering messages in partially-connected ad hoc networks where a connected path from source to destination may not exist. Existing ad hoc routing protocols assume a connected path, but in scenarios with limited connectivity or network partitions, this assumption is invalid. Epidemic Routing ensures eventual message delivery through random pairwise exchanges among mobile hosts, maximizing delivery rate, minimizing latency, and reducing resource consumption. The protocol distributes messages to carriers within connected portions of the network, allowing messages to spread through transitive transmission. It leverages the theory of epidemic algorithms, where random data exchange ensures eventual consistency. Epidemic Routing is implemented in the Monarch simulator, showing that it can deliver 100% of messages with reasonable resource consumption in scenarios where existing protocols fail. The protocol is robust to varying network conditions, including low transmission ranges and limited node mobility. It balances resource consumption and delivery rate by limiting hop count and buffer size. The paper evaluates Epidemic Routing under various scenarios, demonstrating its effectiveness in delivering messages even when traditional routing protocols fail. The protocol is sensitive to node density and transmission coverage, and its performance is influenced by buffer size and message priorities. Future work includes exploring hybrid routing techniques and leveraging positioning information to improve resource utilization. Epidemic Routing is suitable for applications such as mobile sensor networks and disaster recovery, where reliable message delivery is critical despite limited connectivity. The protocol ensures eventual delivery by distributing messages through multiple carriers, making it a viable solution for ad hoc networks with intermittent connectivity.