This article is intended to improve the certainty of the absorbed dose determination for reference dosimetry in CyberKnife beams. The CyberKnife beams do not satisfy some conditions of the standard reference dosimetry protocols because of its unique treatment head structure and beam collimating system. Under the present state of affairs, the reference dosimetry has not been performed under uniform conditions and the beam quality correction factor kQ for an ordinary 6 MV linear accelerator has been temporally substituted for the kQ of the CyberKnife in many sites. Therefore, the reference conditions and kQ as a function of the beam quality index in a new way are required. The dose flatness and the error of dosimeter reading caused by radiation fields and detector size were analyzed to determine the reference conditions. Owing to the absence of beam flattening filter, the dose flatness of the CyberKnife beam was inferior to that of an ordinary 6 MV linear accelerator. And if the absorbed dose is measured with an ionization chamber which has cavity length of 2.4, 1.0 and 0.7 cm in reference dosimetry, the dose at the beam axis for a field of 6.0 cm collimator was underestimated 1.5%, 0.4%, and 0.2% on a calculation. Therefore, the maximum field shaped with a 6.0 cm collimator and ionization chamber which has a cavity length of 1.0 cm or shorter were recommended as the conditions of reference dosimetry. Furthermore, to determine the kQ for the CyberKnife, the realistic energy spectrum of photons and electrons in water was simulated with the BEAMnrc. The absence of beam flattening filter also caused softer photon energy spectrum than that of an ordinary 6 MV linear accelerator. Consequently, the kQ for ionization chambers of a suitable size were determined and tabulated as a function of measurable beam quality indexes in the CyberKnife beam.