The geometric distribution models the number of trials needed to get the first success in a sequence of independent and identically distributed Bernoulli trials. It is characterized by its memoryless property, where the probability of success in future trials is independent of past failures.