An ink jet recording apparatus includes a carriage for mounting therein a recording head which executes a recording operation by discharging ink onto a recording material, a conveying roller for conveying a recording material, and a cap for covering a discharge port of the recording head. In addition, a pump generates a negative pressure in an interior of the cap, a lock member, adapted to be rotated in synchronism with the rotation of the conveying roller, locks movement of the carriage, and a drive transmission mechanism transmits a driving power of the conveying roller to the pump. The drive transmitting mechanism is equipped with a delay mechanism for delaying transmission of the driving power to the pump until the conveying roller has been rotated by an amount corresponding to an amount of rotation of the lock member for shifting the carriage from a locked state to an unlocked state.