I have found these statistics from the Cambridge Structural Database from January 2014 (see below). As you can see, most of the uploaded structures have an R factor between 0.03 and 0.07. If an R factor is "good", strongly depends on the crystal/compound. In some cases (e.g. disorder) a factor of 0.05 can be absolutely alright and in other cases, if the circumstances can be explained, even a structure with an factor of 0.10 can be published (2.2 % of the database). Nevertheless, you should always try to solve and refine the structure to the lowest R factor possible.
R-factor range ; No. in range ; % CSD ; Cumulative %