Hi Sylvain ----

I'd have to test that to be sure but that would fall into the realm of "unanticipated behaviour". If MDOWN() fires on release of the button, you are correct.

>>
>>I have a paper in the KB here about this exact problem. In general, you can use the MDOWN(),MROW(), and MCOL() functions to trap the region where the mouse is being clicked within the Valid and return .T. if it maps out to the Cancel button. Check out the paper, it goes into precise coding.
>
>But then, a user could press (and hold) the mouse on the cancel button, move off the button and release the mouse. This way, the valid didn't fire and the cancel button wasn't pressed either.
>
>Am I wrong on this one?