Although recommender systems have come to give recommendations with high precision, users are not always satisfied with the recommendations. User satisfaction is apparently influenced by many other factors. We specifically examined user intervention as one factor influencing user satisfaction. We tested two hypotheses: user intervention itself improves user satisfaction; and the more users intervene in the recommendation process, the more they are satisfied with the recommendations. We conducted an experiment incorporating user intervention of several kinds to reveal the relation between user intervention and user satisfaction.