[en] The Bureau of Radiological Health and the National Bureau of Standards conducted a nationwide survey of cobalt-60 teletherapy facilities to determine their accuracy in delivery of a prescribed dose to a phantom. Participation was obtained from 684 respondents or 75 percent of US facilities. For each cobalt-60 unit the average dose recorded on five dosimeters was within 5 percent of the prescribed value for 83 percent of the respondents and only exceeded a 10 percent difference for 3 percent of the respondents. Sufficient information for reconstruction of dose calculations was available for 87 percent of the participating facilities. Of these reconstructed calculations, 56 percent were within 1 percent of the prescribed dose of 300 rads. The analysis of data from the survey showed that better performance were related to a number of factors. Among these were expressing the machine calibration in terms of dose rate, performing the machine calibration on an annual basis, and correcting for source decay on a monthly basis. Larger facilities achieved better results than smaller facilities when calculating the absorbed dose rate. Also, each 4 machine characteristics (isocentric mounting, short exposure time, high normalized output, and long treatment distance) were related to better performance with regard to both dosimeter readings and calculated doses. Larger facilities were more likely to be using a cobalt-60 unit having those characteristics