This paper describes modifications to an original design, correction factors and uncertainty evaluations for a graphite pancake ionization chamber constructed at the Institute of Nuclear Energy Research (INER, Taiwan). A bilateral comparison of the absorbed dose to water standards for Co-60 using transfer chambers was performed to verify the graphite chamber experimental accuracy and measurement consistency. The comparison results showed a satisfactory agreement in the measurements within the combined standard uncertainties (k = 1). This paper also compares the absorbed dose to water calibration in medical accelerator photon beams traceable to INER standards following the recommendations given in the AAPM TG-21 and TG-51 dosimetry protocols. For all types of linear accelerators and cylindrical chambers at 13 participating hospitals in Taiwan, the TG-51/TG-21 dose ratios were the same within +/- 1.5%, less than the combined uncertainty, irrespective of the chamber make and model for each photon included here. A quality assurance guide for institutions switching from the TG-21 to TG-51 protocol was suggested based on the comparison results for the two dosimetry protocols.