Saltwater intrusion in coastal aquifers is investigated using stochastic simulation and a numerical groundwater model. Daily rainfall rate was assumed as a stochastic process and using a Transition Probability Matrix model, a set of 100 realizations was generated for a total simulation period of 30 years. These daily rainfall realizations were inputs to a vertical (profile) aquifer model in a transient, variably-saturated and variable-density flow and salt transport simulation. Monte-Carlo analysis for an unconfined aquifer indicates that certain rainfall realizations can produce salinity distributions that strongly deviate from the mean salinity value. Such a dependence of the salinity distribution on stochastic nature of rainfall implies that seawater intrusion prediction could be inadequate if only the historical rainfall timeseries is used as input to the numerical model. This situation could potentially lead to ineffective management plans if such deviation is not taken into consideration.