Online search advertising has become the main stream in online advertisings, and its revenue is up to 40% of the whole on-line advertising market (there are lots of other online advertising such as email advertising, banner advertising). Lots of researches focus on how to increase click through rate for an ad, how to increase the quality of an ad, and how to get better bid stratagem. In this paper, we tackle the problem as how to personalize the scheduling of ads for different users. In fact, we allocate ads to each user based on user history behaviors. At the same time, our personalized schedule algorithm should not decrease search engine's revenue. In other word, our object is that we should delivery the interest ads to individual users based on their interest inferred from user’s history queries and history clicked documents. For each query and document we use five main features such as the unigrams, categories, phrases, brands, loss memory of an ad. These features are extracted from each ad relevant to current query too. Most of these features using a cosine similarity measure to compute a score in order to participate in our personalized user interest model. Our simulation experiments prove that our personalized schedule algorithm is promising on increase the revenue of search engine and the click through rate of each user.