In statistical hypothesis testing, the p-value is the probability of obtaining a value of the test statistic at least as extreme as the one that was actually observe, given that the null hypothesis is true. The fact that p-values are based on this assumption is crucial to their correct interpretation.