Using the net present value is the standard methodology in theoretical analysis, and the most frequently used method for making financial decisions. However, net present value is rarely used in production and inventory decisions. The main reasons appear to be the complexity of the formulae and the robustness of the EOQ model. We investigate the general multiproduct, multistage production and inventory model using the net present value of its total cost as the objective function. A power-of-two heuristic gives us a near optimal solution to this problem. If the base period is fixed (or varied), the solution based on the best power-of-two heuristic will be within 6.2% (or 2.1%) of the optimal. This result is surprisingly similar to models using the long-term average cost. The average cost does not reflect the time value of money. Does this mean that decisions based on average cost are significantly inferior to those based on net present value? The answer is quite surprising. If we include discounted production cost in the holding cost, it turns out that the decision based on average cost is only 9.6% (in terms of the net present value of the total cost) worse than the decision based on the net present value. However, the reorder interval based on the average cost could be much longer than that derived using net present value. This result shows that average cost is a good approximation to the net present value when the demands are deterministic.