Hrvoje Niksic <hniksic(a)srce.hr&gt; writes: I think I have a solution. Do you think that this is evil: - when I create the misc-user-event object (the dnd-data), I cons a reverence to the event itself into it. So the function still has access to it's event. And this is how it looks... events.h defines misc_user_data as: struct misc_user_data { Lisp_Object function; Lisp_Object object; int button; unsigned char modifiers; int x, y; }; event-Xt does the following when a drop occurs (it's a nice reference loop...): XSETEVENT (lisp_event, emacs_event); emacs_event->event.eval.function = Qdragdrop_drop_dispatch; emacs_event->event.eval.object = Fcons ( lisp_event, Fcons (l_type, Fcons (l_dndlist, Qnil))); now event-stream.c catchs the misc_user and does: if (XEVENT (event)->event_type == misc_user_event) { call1 (XEVENT (event)->event.eval.function, XEVENT (event)->event.eval.object); } This calls dragdrop-drop-dispatch which is defined as: (defun dragdrop-drop-dispatch (object) (let ((event (car object)) (data (cdr object))) (message "drop at %d,%d; type %d; data '%s'" (event-x event) (event-y event) (car data) (cadr data)))) And voila: I see the info of the drag in the minibuffer. Is the loop reference evil? It seems to work fine... Regards, Oliver.