Thanks Jim. I guess I have just been lucky to date, my character functions have not had that problem. But I tried your example and I see the problem now. Geez, learn something new every day 'round here!

>The situation is basically that Select will take the length of the *first* field returned and use that length for the field for all subsequent records returned. So that even with a simple thing like ...IIF(x > y, "NO", "YES"), if the first result is a "NO" then any yeses will appear as "YE". So it is always safest to either make sure that any function always returns information of identical length *or* to adjust the returned value (using PADx or 001.000(as many zeroes as needed)*result). In the example above I would code "NO " to cover this.