A sound output device includes a sound generating portion that generates a sound, a sound guiding portion that takes in the sound generated in the sound generating portion from one end, and a holding portion that holds the sound guiding portion in the vicinity of the other end. The holding portion is engaged with an intertragic notch, and supports the sound guiding portion such that a sound output hole of the other end of the sound guiding portion to face a depth side of an ear canal. Even in a state where the listener wears the sound output device, the sound output device does not block an ear cavity of the listener, and the listener can listen to the ambient sound.