During the installation of SQL Server 2012, in the Collation tab, i can choose between French_CI_AS and French_100_CI_AS.

What do they mean and which one is newer? When there's no Code, is it newer or older than the 100?

2 Answers 2


The one without the code is a version 90 (SQL Server 2005) collation, used by previous versions of Windows / SQL Server. The collation including 100 is newer (SQL Server 2008).

There's a bit of an overview of the differences in the SQL Server 2008 documentation, here, however it doesn't go into great detail.

If you're developing a new application, you may find that sorting is slightly more logical (particularly for unusual characters) with the 100 version. However, if you've got any other SQL Servers running with French_CI_AS, it might be better to keep it consistent with them.

  • We havent developed anything yet. Everything is fresh on SQL Server 2012. So i guest the best bet is to use the newest configurtion everywhere. May 3, 2013 at 15:18
  • 3
    CI - Insensitive AS - Accent Sensitive May 3, 2013 at 18:31
  • 1
    I did not find any source that _110 collations exist. Only _80, _90 and _100 families exist (see e.g. msdn.microsoft.com/en-us/library/ms143726.aspx).
    – user50903
    Oct 24, 2014 at 8:13
  • @Cheese, you're absolutely right, as noted in your link the only new collations introduced in 2012 were Supplementary Collations (_SC), and it doesn't look like there's anything new in 2014. I'll update accordingly. Nov 27, 2014 at 3:46
  • I think blank means 80. For example there is Chinese_PRC_90_CI_AI_SC_UTF8, but SC and UTF8 options are not available for blank one.
    – imba-tjd
    Feb 19 at 10:35

SQL server must version the collations. Since collations determine sort order of data persisted in the database the collation must be guaranteed to remain stable between the releases. Otherwise a collation change (eg. fixing a bug in a collation) with a new release of Windows it would result in two rows int he database to sort differently before and after the release. This may sound trivial, but in effect it means is an index corruption because the items should be sorted as A > B (according to the new, fixed, collation rules) but in the database are present as B > A (according to the old, buggy, collation rules). Therefore SQL Server must snapshot the windows collation at each release and stick the release version number in the collation name (since the names do not have an explicit versioning scheme).

The collation you choose is subject to many factors. Consider that if you interact with an older server to exchange data it may not have the newest collation you use.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.