there is nothing special just like many other databases,
oracle supports Unicode, you can define it as default encoding for a database instance you create
when you are there creating connection you have to mention that you want to use UTF-8 or UCS-2 or other unicode  encodings
that can be set through an extra parameter in definition of JDBC connection or connection pool or JNDI resource in your application server or j2ee project.

Generally its possible to save Farsi in oracle
database unless it's character set is "US7ASCII".
But in my experience the best way is to create the
database with "AMERICAN_AMERICA.AR8MSWIN1256"
character set. Be aware that when you create the
database, the default character set is "WE8ISO8859P1",
so you have to choose the character set as mentioned.
Since i don't exactly know that what kind of problem
you have, i can not direct you to the solution. If
setting the character set solves your problem, my
pleasure, otherwise, i am at your service.

Sorry for late reply. Let me put it simple and explain my intention to use oracle  and store farsi  as unicode  in it.
I have a simple web application which must show farsi to the viewer and make it possible to the viewer to search on the data and fill data in farsi.
I'm using jboss as application server, oracle as dbms, seam as web application framework and icefaces.
Any suggestion welcome.

I just forgot to say that i do not know java at all.
I know Oracle database and its 4GL development tools.
What i suggested to Omid was based on this experience.
What i know is that you have to keep both sides,
client and the database, to talk together with the
same character set to prevent code conversion. Oracle
describes that which character sets can be converted
to each other and which can't. If you have two
character sets that can not be converted to each other
then Oracle inserts a '?' for each character of data
which it receives.
Sorry for interruption in java discussions.