The unique characteristics of employee stock options make traditional option pricing models inapplicable to their valuation without substantial modifications. In this thesis, three extensions are made to the standard binomial tree model. First, the dilution effect is factored into the model. Second, instead of computing the option value based on a specific exercise criterion, the pattern of employees’ exercise behavior is explicitly modeled by the chi-square distribution. The third extension is the addition of a state-dependent employee forfeiture rate. This thesis presents comparative analysis of popular models and the models proposed in this article. Finally, the impacts of the modifications on the fair value of employee stock options are investigated. They lend support to the claim that the proposed models are more realistic.