Although many studies have reported a distortion of subjective (internal) time during preparation and execution of actions, it is highly controversial whether actions cause a dilation or compression of time. In the present study, we tested a hypothesis that the previous controversy (dilation vs. compression) partly resulted from a mixture of two types of sensory inputs on which a time length was estimated; some studies asked subjects to measure the time of presentation for a single continuous stimulus (stimulus period, e.g. the duration of a long-lasting visual stimulus on a monitor) while others required estimation of a period without continuous stimulations (no-stimulus period, e.g. an inter-stimulus interval between two flashes). Results of our five experiments supported this hypothesis, showing that action preparation induced a dilation of a stimulus period, whereas a no-stimulus period was not subject to this dilation and sometimes can be compressed by action preparation. Those results provided a new insight into a previous view assuming a uniform dilation or compression of subjective time by actions. Our findings about the distinction between stimulus and no-stimulus periods also might contribute to a resolution of mixed results (action-induced dilation vs. compression) in a previous literature.