Now that the NCBI have stopped messing around with character encodings, HubMed's RIS and BibTeX export should be working properly again. RIS export (eg for Endnote or Procite) replaces UTF-8 characters with their Latin-1 equivalents, while BibTeX export converts UTF-8 characters to their bracketed forms. Still no UTF-8 RIS export (which the newest version of Endnote can handle), but I have an idea for how to add that as an option.