Terry Reigel replied: > Thanks to your testing we now know we did all correctly understand how > it was supposed to work, and that it did change without anyone noticing. Yep, I guess these changes/bugs can happen. The fact that no one noticed before now is probably an indicator of how seldom people merge with custom Flags in the "sending" dataset which do not exist in the "receiving" data set. > ... you can make it work with the flag in the "sending" data set > provided you create one of the same name in the other data set too, > which I think is what you are saying you found works. Yes. If you want to retain the values of a custom Flag in the "sending" data set, all that is required is to create a custom Flag with the exact same name in the "receiving" data set. [Hmmm... I think the names are case insensitive, but should probably retest that. Later...] You do not even have to create the same list of values in the "receiving" data set (although it would probably be wise to do so). The people from the "sending" data set will retain their Flag values even if such values are not in the list of defined values for that Flag in the receiving data set. A bit wierd to see people with values which are not in the defined list and thus cannot be set manually, but it works. > ... that does provide a workable work-around as long as one knows > it is required. Yes, it takes an extra step first to create the custom Flags, but otherwise everything works as one would probably expect. Michael