Suppose that Bernoulli trials with unknown stopping rule terminate in s successes and f failures. For a sensible class of stopping rules, the p-value for a one-sided test is bounded between the negative binomial p-value and twice the binomial p-value for this data.