There are two types of hijacking attacks:
In an active attack, an attacker finds an active session and takes over.
With a passive attack, an attacker hijacks a session, but sits back and watches and records all of the traffic that is being sent forth.
Session hijacking can be active or passive in nature depending on the degree of involvement of the attacker in the attack. The essential difference between an active and passive hijack is that while an active hijack takes over an existing session, a passive attack monitors an ongoing session.
Generally a passive attack uses sniffers on the network allowing the attacker to obtain information such as user id and password so that he can use it later to logon as that user and claim his privileges. Password sniffing is only the simplest attack that can be performed when raw access to a network is obtained. Counters against this attack range from using identification schemes such as one-time password (e.g. skey) to ticketing identification (such as Kerberos). While these may keep sniffing from yielding any productive results, they do not insure the network from an active attack neither as long as the data is neither digitally signed nor encrypted.
In an active attack, the attacker takes over an existing session by either tearing down the connection on one side of the conversation or by actively participating by being the man-in-the-middle. These have been discussed at length under the discussion covering the various steps involved in a session hijack.
This requires the ability to predict the sequence number before the target can respond to the server. Sequence number attacks have become much less likely because OS vendors have changed the way initial sequence numbers are generated. The old way was to add a constant value to the next initial s