Recently I installed the latest official version (5.15) and opened a .bib file. For my surprise, many characters were unreadable. Then I discovered that JabRef is ?detecting? wrongly the encoding as “Shift_JIS” instead of “UTF-8”, despite the file is really saved in UTF-8 encoding by other text editors, like Sublime Text.
First JabRef tries to look for the encoding header and if not present it tries to guess the char set.
Unfortunately, the shift_jis was the first one guessed, utf 8 the second, but both with the same confidence.