An application of stochastic process for describing and analysing daily the rainfall pattern at Universiti Pertanian Malaysia (U.P.M.), Serdang, is presented. A model based on the first-order Markov chain was developed. The model uses historical rainfall data to estimate the Markov transition probabilities. The year is divided into four seasons, each is represented by a separate transition probability matrix. The range of rainfall values is divided into eleven states, thus resulting into 11 x 11 transition probability matrix for each season. The model is capable of simulating a daily rainfall record of any length for the area. It is evaluated by comparing the simulation result with observed data for a one-year period.