Disclosed are a method and an apparatus for calculating similarity of life log data, the method including: producing, by a life log data producing unit, a plurality of life log data on a daily basis, in which at least one estimated activity state is indicated for each predetermined time section, by matching a user's position information per time period and a user's motion state information per time period with an estimated activity table in which the user's estimated activity states are defined in advance; converting, by a modified life log data producing unit, the plurality of life logs data into a plurality of modified life log data which is indicated for each merged time section made by merging a preset number of continuous time sections; and calculating, by a similarity calculating unit, life log similarity among the plurality of modified life log data by comparing the plurality of modified life log data for each merged time section.