A1. When trying the configuration and using sqlldr (utf8) we get quuestion marks in the nvarchar2 column. A2.describe table ; From the original table nvarchar2(50) Dul describe Type 1 varchar2 cs 871(utf8). Setting EXPORT_MODE= FALSE and LDR_OUTPUT_IN_UTF8 = TRUE is the way to go here, as that is the only configuration in which DUL does character set conversion. The idea is that everything is then converted to UTF8 and indicated as such in the sql*ldr control file. Q1: If you do this which one is wrong ? Q2: You can do a describe in DUL of the table, it will then show the columns names and the characterset DUL thinks it is, can you check if that output matches your expectations   yet, instead I have plugged in a relatively new asm i/o in the current production version. Please get a fresh executable from the usual place. I expect that this code can work with asm directly, but the startup perfromance issue has not been addressed (is on the top of my list) I have added support for iw8iso8859p8, new executables are in the usual place. Let me know in case of issues.