I'll take the cue from Karluk, here's what I got from Metalink:
1. Gather the necessary information from your source database. You will need
to know the exact name of all TABLESPACES, so perform the following query
as a DBA user:
SQL> SELECT tablespace_name FROM dba_tablespaces;
2. Perform a full export from the source database as a DBA user. For example:
exp system/manager full=y file=expdat.dmp
3. Move the dump file to the target database server via ftp.
(Note: remember to change to BINARY mode to avoid corrupting the file.
To configure ftp for binary mode, enter 'binary' at the ftp prompt)
4. Create a database on the target server. Create all the tablespaces in your
NT machine similar in size to the one in Unix. The datafiles, of course
are different.
5. Before performing the import, you will need to precreate your tablespaces.
This is necessary since the import will want to create the corresponding
datafiles in the same file structure as was at the source database. Since
your file structure will be different on the target database, precreating
the tablespaces will allow you to specify a file structure that will work.
6. Perform a full import with the parameter IGNORE=Y into the target database
as a DBA user.
imp system/manager full=y ignore=y file=expdat.dmp
Using IGNORE=Y will tell Oracle to ignore any creation errors during the
import, allowing the import to complete.
Can not ellaborate more on how you'll gonna create the DB bec. I assume that you already know the steps.