In search advertisements, advertisers have to distribute the budget for an advertising campaign to a series of sequential temporal slots (e.g., daily budgets) during a certain promotion period. However, advertisers can not know exactly the optimal budget for each temporal slot because there exist many kinds of uncertainties in search advertising markets. In this paper we presented a stochastic model for budget distribution over a series of sequential temporal slots during a promotion period, for a given campaign, considering the optimal budget for temporal slots as a random variable. We studied some properties of our budget decision model. We also made some experiments to evaluate our model, and experimental results showed that our model could outperform a baseline strategy commonly used in practical search advertising campaigns.