This note discusses some problems possibly arising when approximating via Monte-Carlo simulations the distributions of goodness-of-fit test statistics based on the empirical distribution function. We argue that failing to reestimate unknown parameters on each simulated Monte-Carlo sample - and...