An image acquisition method in which a pulsed illumination beam emitted from a light source is scanned while being focused at a sample, signal light generated as a result of a non-linear optical process at each scanning position is detected, and an image of the sample is generated on a basis of the detected signal light, the image acquisition method including: acquiring a mixed image, which includes in-focus signal light generated at a focal position of the illumination beam in the sample and which also includes out-of-focus signal light; acquiring an image of the out-of-focus signal light on a basis of a plurality of mixed images having mutually different intensities of the out-of-focus signal light; and acquiring an image of the in-focus signal light by subtracting the image of the out-of-focus signal light acquired, from the mixed image acquired.