>>>Have you seen this Bonnie?
>>>http://www.sql-server-performance.com/datatypes.asp
>>>
>>
>>Many written things are mistaken.<
>
>Are you agreeing with me that it's contradictory? Are you saying it's wrong? Do you have any information to answer my original question?
>
>~~Bonnie

I don't read all, but one glanced at has been enough to me;
I uses this simple rule:
if it demonize the NULL, then i stop the reading.
If one writes this on MS SQL,
then it means that it has understood "NULL".

I have not made many tests, but i think that is difficult to confront a varchar with a char of similar dimensions and finding one meaningful difference of performances.
Normally the data size to move is more important of the computation complexity
(why in the normal CPU the cache it is much small and when the
RAM is used the speed down of 10-50 times ( seek index !! ), worse for the HD);
the complexity of a VARCHAR is much small: its length.

Therefore I suggest this rule:
If the medium length is > 90% and the variance is not excessive use a CHAR, otherwise use a VARCHAR (attention, I do not use character keys !).
Of course, if it tightens it it must frequently concat with others,
and it must to be RTRIMmed, then use always a VARCHAR.

Pratically: i uses CHAR(N) for N<6

Fabio