The sign and size of estimates of the elasticity of currency substitution critically depend on the definition of the oppurtunity costs of holding money. We investigate possible biases by means of Monte Carlo experiments, as sufficient real data are not available.