>>That's brilliant, thanks. You are right thought just being able to say which level back you want to go to would be perfect rather than having to parse the procedure name out.
>
>I'd avoid this like the plague - you're blowing the call stack and one or more execution stacks out without being aware of the consequences of them. It's an open invitation to having dangling object refs left lying about, since there may be methods bypassed by RETURN TO that never terminate, and since the code in them is active, you'll likely not be able to release the objects involved (and their parent containers, etc.) at a later point.

I understand what you are saying but this is for one exceptional case where a custom class (to talk to word) is called. If it crashes I don't want any of the remaining code to run just the Destroy (which appear to run happily) then for control to return to the button that was pressed to start it.