Purpose: To analyze absorbed dose calibration coefficients, ND,w, measured at accredited dosimetry calibration laboratories (ADCLs) for client ionization chambers to study (i) variability among ND,w coefficients for chambers of the same type calibrated at each ADCL to investigate ion chamber volume fluctuations and chamber manufacturing tolerances; (ii) equivalency of ion chamber calibration coefficients measured at different ADCLs by intercomparing ND,w coefficients for chambers of the same type; and (iii) the long-term stability of ND,w coefficients for different chamber types by investigating repeated chamber calibrations. Methods: Large samples of ND,w coefficients for several chamber types measured over the time period between 1998 and 2014 were obtained from the three ADCLs operating in the United States. These are analyzed using various graphical and numerical statistical tests for the four chamber types with the largest samples of calibration coefficients to investigate (i) and (ii) above. Ratios of calibration coefficients for the same chamber, typically obtained two years apart, are calculated to investigate (iii) above and chambers with standard deviations of old/new ratios less than 0.3% meet stability requirements for accurate reference dosimetry recommended in dosimetry protocols. Results: It is found that ND,w coefficients for a given chamber type compared among different ADCLs may arise from differing probability distributions potentially due to slight differences in calibration procedures and/or the transfer of the primary standard. However, average ND,w coefficients from different ADCLs for given chamber types are very close with percent differences generally less than 0.2% for Farmer-type chambers and are well within reported uncertainties. Conclusions: The close agreement among calibrations performed at different ADCLs reaffirms the Calibration Laboratory Accreditation Subcommittee process of ensuring ADCL conformance with National Institute of Standards and Technology standards. This study shows that ND,wcoefficients measured at different ADCLs are statistically equivalent, especially considering reasonable uncertainties. This analysis of ND,w coefficients also allows identification of chamber types that can be considered stable enough for accurate reference dosimetry.