An image forming apparatus that forms an image by scanning a recording medium with a plurality of light beams that are modulated by individual image forming signals respectively corresponding to the light beams. A beam detecting device detects the plurality of light beams at preset positions and outputs corresponding detection signals. At least one delay device delays each of the detection signals from the beam detecting device by an arbitrary time period which is predetermined individually for each of the detection signals. A reference clock signal generating device generates a plurality of reference clock signals on the basis of delayed detection signals output from the at least one delay device. The image forming apparatus performs an image forming operation according to the plurality of reference clock signals.