Location-based advertising or application has been one of the drivers of third-generation mobile operators marketing efforts in the post few years. As a result, many studies on location-based marketing or advertising have been proposed for recent several years. However, these approaches have two common shortcomings. First, most of them just suggested the theoretical architectures, which were too abstract to apply it to the real-world cases. Second, many of these approaches only consider service provider (seller) rather than customers (buyers). Thus, the prior approaches fit to the automated sales or advertising rather than the implementation of CRM. To mitigate these limitations, this study present a novel advertisement recommendation model for mobile users. We call our model MAR-CF (Mobile Advertisement Recommender using Collaborative Filtering). Our proposed model is based on traditional CF algorithm, but we adopt the multi-dimensional personalization model to conventional CF for enabling location-based advertising for mobile users. Thus, MAR-CF is designed to make recommendation results for mobile users by considering location, time, and needs type, To validate the usefulness of our recommendation model, we collect the real-world data for mobile advertisements, and perform an empirical validation. Experimental results show that MAR-CF generates more accurate prediction results than other comparative models.