The end of a decade is a traditional, even if somewhat arbitrary, time for looking back at what has been achieved, and for looking forward to what may come in the next decade. In the past, epochs of machine translation (MT) began and ended at the midpoints of decades. In the mid-1940s, Booth and Weaver first talked about the use of the newly invented electronic computers for translating natural languages; in the mid-1950s, a MT demonstration by IBM and Georgetown University prompted the start of large-scale official support in the United States and the Soviet Union; in the mid-1960s, the notorious ALPAC report appeared which effectively brought to an end projects in the United States and influenced MT funding throughout the world; and in the mid-1970s, the revival of MT began with increasing operational installations, commercial development and expansion of research activities. However, most of this revival has taken place during the 1980s. This is a general review of the last decade with some suggestions of what might or ought to happen in the next. It is neither comprehensive nor detailed, and full references to the systems mentioned are not given: these are to be found in Hutchins (1986, 1988), and proceedings of conferences in 1988 and 1989 (Carnegie Mellon University 1988, Maxwell et al. 1988, Coling 1988, MT Summit 1989).