IP provides some interesting options, one of which is forced routing. If forced routing is employed, the route to follow is specified in the options field. Padding is used to fill out the last field so that the data field starts on an even 32-bit boundary.

Routers use the source and destination fields to send the packet to the next router. The Internet uses a best-effort, hop-by-hop routing system to keep the handling of the packets simple. Much is made of the fact that the Internet is a CLNS and therefore can route packets almost arbitrarily and those packets arrive out of order or not at all. Although it is theoretically true that a router could choose a different path for the next packet with the same destination address as the previous packet, let us consider the practical realities. First, if the chosen route is good for the current packet, the chances are excellent that it is also good for subsequent packets. Second, the router normally has a limited number of entries in his routing table: the normal route, the secondary route, and the default gateway. Therefore, the number of different paths that the packet could theoretically take is large. In practice, it is relatively small. The practical result is that most packets follow the same path and generally arrive in order, but there are no guarantees.