Association rules are useful for determining correlations between items and have applications in marketing, financial and retail sectors. Lots of algorithms have been proposed for finding the association rules in databases. Most of these algorithms treat each item as uniformity. However, in real applications, the user may have more interest in the rules that contain those fashionable items that occur frequently. Usually too many outdated items exist in databases, but they seldom occur recently. Those outdated items hamper us to find the interesting rules efficiently and effectively. Another case is the user sometimes may want to mine the association rules with more emphasis on some items. To solve these problems, in this paper, we propose the vertical and mixed weighted association rules. We can divide the database into several time intervals, and assign a weight for each interval. Furthermore, we also assign a weight for each item to identify the important items. We present an algorithm MWAR (Mixed Weighted Association Rules) to handle the problem of mining mixed weighted association rules. The experiments show that the rules from our methods have much better predictive ability on future data. We also demonstrate the efficiency of our methods on real data and synthetic datasets.