An endoscope system includes a light source configured to generate either one of first illumination light and second illumination light, an image pickup apparatus configured to obtain an image by picking up an image of an object, and a processor. The processor is configured to make an instruction to set an observation mode of the endoscope system at either one of a first observation mode and a second observation mode, detect an observation state at a time of observing the object based on a feature value calculated from pixel values included in pixels forming a first image, and invalidate an instruction to set the observation mode of the endoscope system at the second observation mode when the observation state at the time of observing the object is not an appropriate observation state.