The accumulation of experience in the application of linear programming and its extensions by farm management workers, together with the growth of the farm advisory profession, should ultimately result in a fruitful interaction. However, this attractive prospect should not be allowed to mask the fact that to date most Australian uses of programming in agriculture have been basically research oriented. To do this it has often been necessary to abstract from those unique characteristics of a particular farm situation which would severely detract from study of general principles and policy. While there is no denying the usefulness of such generalized studies it should be recognized that farm investigations for advisory purposes may require more of a "warts and all" approach. A problem which could typically be involved in such an approach arises when the optimum location of various enterprises is not determinable by orthodox programming procedures because of the lack of coincidence of soil-type and paddock boundaries. A procedure which should, in general, enable determination of a near optimum solution in such cases is presented.