An image forming apparatus monitors sleep-cancelling events, and when a sleep-cancelling event occurs, stores an identifier and a sleep-cancellation time of the event. If a start time of a sleep mode is reached when (i) the apparatus is in a standby mode and (ii) one or more sleep-cancelling events specifying sleep-cancellation times later than the start time have been stored, the apparatus calculates power consumption required to maintain the standby mode for a period from the start time to the earliest sleep-cancellation time, as a power-saving amount savable during the period. Also, a recovery power amount required for recovery from the sleep mode to the standby mode is calculated on the assumption that the sleep mode is started at the start time and is cancelled at the earliest sleep-cancellation time. Switching to the sleep mode is performed only when the power-saving amount is greater than the recovery power amount.