Most individuals covered by Social Security are currently entitled to virtually complete coverage of hemodialysis and/or transplantation treatments. The level of expenditure required for such a program obligates planners to carefully estimate future demand. This paper demonstrates three methods for projecting future need for hemodialysis beds. Two methods are deterministic and project only average future needs. These average needs are first calculated by solving a series of deterministic recursive difference equations and the results are then reaffirmed by a Markov chain analysis. The probabilistic forecasting is handled by repeatedly simulating the possible flows in the patient system. A sensitivity analysis has been carried out and final projections are heavily influenced by the percentage of patients receiving home dialysis treatment. Although specific assumptions are employed, the methods described allow one to use figures or assumptions most suitable for a particular setting. In addition, the methods and approaches described are quite general and thus are applicable to many other types of patient services or dynamic medical systems.