In smart homes, user-aware device usage detection is one of the fundamental tasks. User identification methods with no burden to users have been proposed. However, these methods rely on camera images, which have privacy issues for in-home scenarios. In this paper, we present a user identification method via a touch-screen button operation. The key idea is to utilize users’ habits of button operations to identify users. We extract features from a time series of touch-screen operation data and identify users using supervised learning. Our experimental evaluations demonstrated that our user identification method identified users with an accuracy of 94.4%. With the limited amount of training data obtained in 10 trials, the accuracy was 92.8% when we used the latest training data, confirming the feasibility of our user identification method.