Hrvoje> No. Isearch uses the overriding local map because it Hrvoje> wants to take precedence over extent-local keymaps. Hmm, OK, that's a good reason. Hrvoje> I wouldn't do this because some parts of isearch might Hrvoje> depend on overriding-local-map being used. Hrvoje> Why not do this: Hrvoje> (set-keymap-parent folding-isearch-mode-map overriding-local-map) Hrvoje> (setq overriding-local-map folding-isearch-mode-map) Thanks for the feedback. I tried this. Entering isearch, then entering folds works fine. What breaks is exiting isearch (by pressing Enter, C-a, C-g, etc.). It sort of gets out of isearch, so you can move around the buffer and such, but then when you press a key bound to self-insert command, you end up back in the original isearch. So, say you did "C-s g n u Enter", move around the buffer, then press "s", and you see isearch prompt saying "gnus." My guess is that this is related to the following function which wraps around everything in when doing isearch in folding mode. ;; The HEART! Executes command and updates the foldings. ;; This is capable of detecting a `quit'. (defun folding-isearch-general (function) ... (cond ((memq function '(isearch-abort isearch-quit)) (setq quit-isearch t)) (t ... (funcall function) ... (if quit-isearch (signal 'quit nil)))) So, Enter is bound to isearch-exit would follow somewhat different code path than isearch-abort ("C-g"), but both end up having the same problem. There is some kind of cleanup which is not happening. I just could not figure out what it was. I even tried sticking in (isearch-done) for "C-g" (which should set overriding-local-map to nil), but I kept having the same problem. (if quit-isearch + (isearch-done) (signal 'quit nil)))) Gleb