The rather "limited" utf8 support in tcc wouldn't be so bad if you could force a conversion on input/output and otherwise treat stings with the default non-utf codepage. At least @UTF8ENCODE[s,...] works fine, but %@UTF8DECODE[s,...] only returns ? chars whatever code page I've set. The manual reads: "@UTF8DECODE[s,string] : Create a text string (using the current code page) from a UTF8 input string. Returns the text string." But if I run the attached utf8-encoded batch file containing a test string I get Active code page: 1252 Test??????? Active code page: 850 Test??????? Active code page: 65001 Test??????? Active code page: 437 Test??????? How is @UTF8DECODE[s,...] supposed work, am I missing something here?