Our research into the problem of scheduling in distributed computing systems indi­cates that several techniques and tools from the area of "expert systems" can be successfully adapted for use in the design of a smart distributed scheduler. In this paper we look at expert system approaches to the representation of imprecise knowledge, techniques for reasoning about imprecise and unreliable knowledge, and means for handling knowledge accumulated over time. We show that that these are precisely the types of knowledge a distributed systems scheduler must deal with in order to make scheduling decisions and we give examples of the use of these techniques in the realm of task assignment and task migration algorithms. We then describe a task migration algorithm we have designed which utilizes rule based programming and expert systems techniques to deal with out-of-date and thus potentially unreliable data in system load tables.