please help me with a unicode problem.

I have SQL script that insert data into oracle9i database table.
Database is run in utf8 character set. It sends back correct umlaut symbols, and stores them correctly when i launch my SQL script saved in utf8 using SQLPLUS on windws 2000 mahcine.

But. When I save this SQL into CVS on a redhat9 server which locae is en_US.UTF8 , them do cvs update, and call
sqlplus sfgsg/dgfdg@fdgdf @my.sql

the data is stored incorrectly
for example letter is stored like C5 or something like htis.

Please help me to find the cause of this problem.
I have writeen here all what i know about it.