The main objective of this work is to test empirically the efficiency and the accuracy of the Least Squares Monte Carlo Approach (LSM) in the valuation of projects involving real options. Initially, a series of sensibility tests will be used for its validation. After this, some examples of projects of petroleum exploration and production (E and P) with real options will be elaborated, and their values determined by LSM. The results will be compared with those obtained using the binomial model. It was chosen as benchmark to analyze the efficiency of the LSM due to its simplicity and wide use. Due to similarities between investments in financial and real assets, many studies made with financial instruments have been adapted for economic valuation of projects. Many researches about real options were developed in exploration of natural resources, especially in petroleum E&P. This happens due to the volume of the investments involved and to its peculiar characteristics: the petroleum market is quite developed (presence of future market, financial instruments of protection, derivatives etc); the investments happen to be in an atmosphere of economical and/or technical uncertainties; the projects demand a series of managerial flexibilities (alternative periods for execution of the investments, possibility of changes in the scale of the project etc). Such characteristics require a more careful evaluation of these real assets. A new tool developed in this sense is the LSM. The method consists of the valuation of american options through simulations and simple regressions.