That is the chicken and egg problem I was talking about. When the publication list sort order is 'by citation order', and 'only show cited publications' (as per ams), then \cite looks at the list to find the number, and the list checks the \cite order for what to include. All is good if the ref is actually there, but intercepting errors is problematic.
I see ... just curious: do you happen to know how the LaTeX folks handle wrong citation keys?
Wouldn't it be nice to just replace the "(Xxxx, 0000)" entry by "??" to make it consistent with the cross referencing errors?
I did it this way, because I prefer better-matching page breaks over visual similarity. At least some of the other \cite options generate ?? instead.
I'm afraid I don't follow you ... what does it have to do with page breaks? Oliver