Which table is the textbox bound to? The parent or the child? Do you change the area that is selected in the Valids? If the select is changed from the parent and the record pointer moved in the child then the relation is suspended until you change the record pointer in the parent again.

P.S. The valid is kind of a strange thing to be calling to change the colors. JMO.

>>David-
>>
>>Is there only one table involved? More? Relations? A parameterized view? Is the errant textbox showing the next or last record?
>
>There are two tables. The work order table, an operation table (two field table that maps operation codes to operation descriptions). The work order table is related to the operation table with an opcode field. I use this in a combolist to allow the user to select from a list of options that may apply to a work order.
>
>No parameterized view.
>
>The errant textbox seems to be showing the next record (where next is determined by the index order of the table).
>
>Everthing about the tables is set up in the DE except that I do not autoopen the tables. This is so that I can specify the database to be used at run time.
>In the LOAD I set the table's cursor's DB to my.dbc then fire the DE.opentables.
>
>>
>>What is your refresh code?

>LOCAL lcproc
>lcproc = SET("procedure")
>SET PROCEDURE TO UTILS.PRG ADDITIVE    &&contains ibox udf
>
>dodefault()
>
>THISFORM.txtqtyinproc.valid           &&set color of txtbox depending on value
>THISFORM.txtihcdate.valid
>
>IF alltrim(THISFORM.txtcomment.value) <> alltrim(wojrn.comment)
>	ibox("Houston, we have a problem")
>ENDIF
>
>SET PROCEDURE TO &lcproc
>

>
>
>
>
>
>
>
>
>>
>>Is the rowsource for the CBO the same as the source for the textboxes?
>>
>>>One other clue could be that I do not refresh in the combolist's valid. Only in it's interactivechange. I've gotten a little hinky about refreshing in the valid of a combolist after some of the stuff I've read. Again, everything does appear to be refreshed at the right times, except the one field, when this problem occurs, so it's hard to believe lack of refresh is a source of the problem.
>>>
>>>I'm thinking some kind of server delay/buffering interface/race condition weirdness could be at fault...explaining why selecting from the combo's dropdown list (rather than using the interactivesearch) clears the problem.