As Brennan and Schwartz (1985) point out, the valuation of natural resources projects is particularly difficult due to the high degree of uncertainty in output prices of resources. In general, there are two competing procedures to price risky projects in natural resources developments. One is decision analytic, based on stochastic dynamic programming (Smith and McCardle, 1998), and the other is contingent claims analysis, based on the no-arbitrage theory of financial markets (Brennan and Schwartz 1985, Paddock et al. 1988, Laughton 1998, Schwartz 1998). In this paper, we use the second approach to develop a new model, and the major developments in this study are as follows. First, by considering two fundamental stochastic variables, output prices and development costs, we develop a pricing model more realistic than most of the previous models, which use a single random variable, such as an output price. Second, we test the validity of our model using real data. Test results reveal that our model is easily applicable to the real world in the capital budgeting decision, implying that the model is more useful than the traditional NPV (Net Present Value) method in the decision of a natural resources investment.