A burden estimation device includes a state space configuration unit, a history learning unit acquiring from the user and storing in a storage unit data related to the burden on the user when the user travels to the destination along the travel route, an individual burden estimation unit evaluating a total travel burden and estimating an individual burden that is the burden on the user in correspondence with each of the states included in the state space through reinforcement learning that sets the total travel burden as a reward, and a total travel burden estimation unit estimating the total travel burden corresponding to an unknown travel route based on the estimated individual burden if the unknown travel route is set and the individual burden has already been estimated for at least some of the states defined in correspondence with the unknown travel route.