Hi, It seems that omniORB adds one or more atexit() handlers. They either abort or experience an uncaught exception when exit() is called after ORB::destroy(). The workaround is to call _exit() after ORB::destroy(). This is a bug as far as I can tell. I am seeing this on Solaris 8 / SparcWorks 5 and also on RedHat 7.0 / gcc 2.96. Anyone else have any insight? Regards, -Huw Rogers