Although a virus contains several epitopes that can be recognized by cytotoxic T lymphocytes (CTL), the immune responses against different epitopes are not uniform. Only a few CTLs (sometimes just one) will be immunodominant. Mutation of epitopes has been recognized as an important mechanism of immunodominance. Previous research has studied the influences of sporadic, discrete mutation events. In this work, we introduce a bounded noise term to account for the intrinsic stochastic nature of mutation. Monte Carlo simulations of the stochastic model show abounding complex phenomena, and patterns observed from the numerical simulations shed lights on long term trends of immunodomnance.