This study contributes to the valuation of employee stock options (ESO) in two ways: First, a new pricing model is presented, admitting a major part of calculations to be solved in closed form. Designed with a focus on good replication of empirics, the model fits with publicly observable exercise characteristics better than earlier models. In particular, it is able to account for the correlation of the time of exercise and the stock price at exercise, suspected of being crucial for the option value. The impact of correlation is weak, however, whereas cancellations play a central role. The second contribution of this paper is an examination to what extent the ESO pricing method of SFAS 123 is subject to discretion of the accountant. Given my model were true, the SFAS price would be a good proxy. Yet, outside shareholders usually cannot observe one of the SFAS input parameters. On behalf of an example I show that there is wide latitude left to the accountant.