In case anyone is still not clear on why quoting lambda expressions is a good thing, consider the following snippet of code: (add-hook 'foo-hook (lambda () (let ((i 1)) (setq foo-map (make-sparse-keymap)) (while (< i 4) (define-key foo-map (intern (format "button-%d" i)) nil) (incf i))))) (add-hook 'foo1-hook '(lambda () (let ((i 1)) (setq foo-map (make-sparse-keymap)) (while (< i 4) (define-key foo-map (intern (format "button-%d" i)) nil) (incf i))))) After bytecoompiling and loading do: foo-hook => (#<compiled-function nil "...(43)" [1 i make-sparse-keymap foo-map 4 define-key intern format "button-%d" nil] 7>) foo1-hook => ((lambda nil (let ((i 1)) (setq foo-map (make-sparse-keymap)) (while (< i 4) (define-key foo-map (intern (format "button-%d" i)) nil) (incf i))))) Note that quoting lambda prevents the bytecompiler from bytecompiling it. Hrvoje Niksic <hniksic(a)srce.hr&gt; writes: I'm not sure either, but I know that it is inconvenient attempting to identify where an error came from in a bytecompiled lambda expression on a hook. One point being you can't just evaluate the hook variable in *scratch*, modify something and reevalute it -- it'll puke on the un(lisp)readable bytecode object. Yup. Just what is gdb-highlight doing in NEWS anyway? Or do I want to know?