>Logic would say that you are absolutely correct but I tried recreating that checkbox (again) but still no luck. The only thing different that I know of is that the first 102 checkboxes were created at the same time. Therefore, I was able to put the names of the fields to which the checkboxes are bound in alphabetic order in the table. The errant checkbox (#103) and one other (#104) were added a couple of days later so their fields were tacked on at the end of the table structure. AFAIK, this should make no difference (although learning that I'm wrong wouldn't surprise me at this point <g>.) The other late-addition checkbox/field combination behaves properly.
>
>For now, I guess I just have to accept the fact that I must test for type before I evaluate the value. It's a pain but at least it's only one checkbox, not all 104!

There's nothing special about the fieldname involved, or the object name? Those are the only other 2 things I think that could have an impact.