Like zip codes and telephone numbers, IP addresses are assigned in a hierarchical fashion. Unlike zip codes and telephone numbers, IP addresses are designed to be interpreted by computers, which operate on a binary numbering system, so IP addresses are organized into a binary hierarchy.

Therefore, to understand IP addressing, it is necessary to convert IP addresses into binary, as in this example with the address 128.8.74.1. Each of the four components in the dotted decimal IP address is converted into eight binary bits.

The first sixteen bits, corresponding to the 128.8 portion of the address, are a unique prefix identifying the University of Maryland. A router outside of the University would require only a single routing table entry, indicating that any address beginning with the sixteen bits 10000000 00001000 (in decimal, 128.8) should be sent towards UMD.

Having reached the University of Maryland, a further routing decision must be made, using some or all of the remaining bits. In this case, a routing table entry matches the first twenty four bits 10000000 00001000 01001010 (128.8.74 decimal) and directs the packet towards Worchester Hall.

The router at Worchester Hall discovers that one of its Ethernet interfaces (for example), is configured to match the twenty four bit prefix 10000000 00001000 01001010. At this point, the layer three routing process is complete, since the packet is now in the hands of a router with an interface configured to match the destination address. The final destination should be directly reachable via that interface, so the router relies on a layer two protocol such as ARP to resolve the IP address into an Ethernet address and complete delivery.