Expanding on what kapilraj already explained:

A route is basically a rule "send everything destined to network <destination> to IP-address <gateway>". This gateway can be one of your network interfaces or the address of a remote machine. In the first case you tell your own machine how to handle a (class of) network packet(s) in the latter you rely on the remote machine knowing how to further dispatch your packets. Usually this is a router whichs purpose is exactly that.

A special case of a route is the "default route", which is used by your machine as a last resort - "if everything else doesn't apply follow this rule".

In a machine with one NIC there are usually two routes: The first one tells the machine to send everything destined to the "own network" (the network the own IP-address is part of) to the own NIC, the other - default - route tells the machine to send everything else to the next router (which hopefully knows how to deal with it).

In a machine with two (or more) interfaces you set up routes similar to the first one in the last example for every interface. At last you set up a default route to one of the routers on one of the networks connected to that way. Every traffice not addressed to the networks your machine already has a leg in will go there.

Of course you can set up routes for specific networks and this way enforce that some traffic goes over a specific interface. The easiest way to get the idea is to draw a (however simple) picture with the networks and the routers and visualize your rules before implementing them.

bakunin