The vocabulary compiled and currently in use at "FIAT" comprises approximately 4000 words and is continually increasing. Some words are uniterms and others are polyterms selected during the scanning of documents of specific interest to the company. An attempt was made to retain a link with natural language by treating proper names and functors in a special way. Although the keywords contained in the vocabulary are mainly in Italian, other languages are also represented. The foreign terms adopted are those representing concepts which the Italian language does not yet cover. The vocabulary proved to be comprehensible to persons whose native language was not Italian, such as trainees of various nationalities who were employed by "FIAT." An experiment was carried out also with bibliographical documentary tapes available on the market. This experiment proved to be somewhat less successful mainly because instructions for use of the thesaurus were different for input and output. The guiding principles for future work are summed up as follows: deeper linguistic exploration of the problem, more assistance for users, and definition of the role of the thesaurus with batch or interactive processing. (Author/AMH)