It seems like the LabVIEW developers have built some features into LabVIEW to protect against the memory leaks and other problems which often result from not closing ActiveX refnums. The refnums now seem to be closed when the application completes execution. This should only be a problem for you if you are letting the application complete execution and then try running it again. If the application is still active, the refnums should remain open. For example, if you take your VIs diagram and drop it ia a loop such that it executes twice, it behaves the way you expected it.