The increased production and publication of professional and scientific literature makes it necessary that abstracts are pro duced in a quick, efficient and economical way. This can be achieved by the mechanization of abstracting. With the aid of computers, extracts can be produced of all kinds of texts which are available in machine-readable form. The main problem of this procedure is how to determine the key sentences of a text, i.e., the passages that contain the most relevant information. Various methods have been developed for this purpose; the one presented here is based on the fact that in order to convey relevant information, subject terminology is used. In many cases subject terminologies are now available in machine-reada ble form too and thus can be easily applied to the automatic production of abstracts.