We provide a framework for testing the probability of ruin in the classical (compound Poisson) risk process. It also includes some procedures for assessing and comparing the performance between the bootstrap test and the test using asymptotic normality.