The purpose of this study was to develop a probabilistic model that could be used by petroleum refiners to evaluate the economic potential of refinery capital investment proposals. The following two requirements were placed on the development at the outset: (1) that the model use linear programming to simulate refinery operations; and (2) that the model keep computer time within reasonable bounds. A probabilistic model was developed that requires the following steps for its application: (1) use linear programming to simulate both the operations of the existing refinery and the operations assuming that the investment is made; (2) select two variables that can be treated as probabilistic variables and assign either a theoretical or a subjective probability distribution to represent future values for the two variables; (3) develop return on investment interpolation data by computing a return on investment for all pair combinations of three tenth year values for each of the two probabilistic variables; (4) develop a return on investment distribution by selecting values at random from the two probability distributions and interpolating among the interpolation data to obtain return on investment data; (5) interpret the return on investment distribution. The model was applied to an actual refining situation …