The rapid growth of location-based social networks through attracting millions users, reveal the high popularity in the short time period. A result of the location based social network services, access to a large collection of data that can extract the spatial history, social relationships structure, movement behavior and characteristics of users. Social, spatial and temporal data analysis leads to create a wide range of location based services. By using statistical and knowledge discovery techniques for advising unvisited locations to users and decreasing large data volume problems, recommendation systems have become the popular services of these networks. Recommendation system is an approach to deal with the problems caused by large amount and growing volume of information and helps user to approach his/her goal among the enormous amount of information. In this paper, we design the novel GEO-FIF method to recommend unvisited places to tourists based on their location history. This time, location and preference aware recommendation system, offer a set of locations near to the user's current position with regard to the time term and the geographical distance between end user and other users as well user preferences that is extracted from visited locations. In the first stage, we survey the impact of the distance between users on common visited locations. Then we measure users’ interest into the locations by creating a user – location matrix within the period of one day and utilizing the content based filtering that for each user calculates a score for each place. In the next step by having the current position of the user, collection of places is limited according to their distance from the user. This is in addition to speed up processing and computing, increase recommendation accuracy. Finally, by using an innovative function, we estimate the similarity of the target user and other users based on the combination of the distance between them and the given score to places by other users. The score of a place for the target user is calculated by accumulative scores given by other users. In fact, we utilize collaborative filtering method to measure the similarity between users and predict the user interest to a new location based on accumulative scores of similar users. Finally combining these methods provides k unvisited locations within specific distance to user’s real-time location and its current time period. In this paper, Gowalla check-in data for Beijing, China were used in the period between October 2011 and November 2011. To evaluate the performance of the proposed method, the results of this method were compared with the results of two basic recommendation system methods in terms of rank accuracy indicator which is most common method for assessing recommender systems. The proposed method increases precision 15 and 12 percent in compared to user-based collaborative filtering using binary methods and GM-FCF, respectively.