One of my previous editorials (AJNR 2008; 29: 1805) dealt with the issue of open access (OA) and its influence on medical journals. Although complete OA is laudable, it is a difficult economic model to sustain. The BMJ was one of the major first OA journals and also one of the first to return to a subscriber fee model. In 1996, the Journal of Clinical Investigation became OA and starting this year it returned to a subscriber model (http://scholarlykitchen.sspnet.org/2009/02/26/end-of-free-access/). Both, BMJ and JCI are prestigious, widely cited journals initially perceived as ideal models for OA. Moreover, JCI charges authors for submissions, number of pages, illustrations (extra for color) and supplemental data (of these, AJNR only charges for color). During its OA period, JCI lost 40% of its institutional subscribers (they are very important to all journals because the fee for this service is higher than that for individuals). This is chiefly due to the fact that JCI does not contain enough non- OA material to justify paying for subscriptions; it contains only research articles all which are OA. AJNR offers enough non-OA contents that offer extra value to its subscribers. At the end of last year AJNR proposed a pay for OA option. Taking into account all OA options, the January and February issues of AJNR published 81 articles, 24 of which were OA. Since February 1^st, 70 articles appear in Publication Preview and 35 are OA! I think that if AJNR offers 50% of contents on an OA basis and the other one half on a subscription basis we will have achieved a nice balance without endangering our economic well being. The pay for OA option has been more popular than I anticipated and as our advertisements decrease, it will serve to offset those declining revenues. OA is here to stay but perhaps only in limited form. While our goal continues to be the widest possible dissemination of science while maintaining a high quality publication, electronic and print publishing are expensive and I do not believe that complete OA would serve AJNR best.