SQL Server Collation

Apart from backward compatibility concerns that you can't solve or work around, there is no good reason to use the first two collations over the third (10.0 is a later version and has more Unicode mappings, and BIN2 uses the fully correct Unicode sorting).  The question is between the third and the fourth. That's a UTF8 question. If you don't need to use UTF8 in (var)char fields (i.e. if you store Unicode only in n(var)char fields, as UTF-16), then the drawback of the fourth collation (vs the third) is that the fourth collation is going to be slower on certain operations on (var)char fields, perhaps substantially so, but I haven't seen tests on that.