In European-funded project MIS under the MLIS programme, the authors attempted a computer-driven translation package for tourism texts in 5 languages. It was believed such a package would be possible due to the highly formulaic language of tourism brochures and business communication in this sector, which ought to allow translation equivalences at the phrase or sentence level to be identified and used. While this proved to be broadly true, the multilanguage format produced errors of agreement and ordering normally avoided by human and computer translators working between two languages. Even at the phrase and sentence level problems of interlanguage equivalence persisted. Awareness of these problems should aid the planning of translation packages of this kind so errors can be avoided.