The author has measured the exposure rate of 56 orthovolt X-ray units. A comparison of consecutive measurements resulted in a 4.2% standard deviation which is mainly due to output variation of the X-ray equipment itself and due to the variation of mains supply. The repeatability of an irradiation controlled by a timer is therefore +/- 9% (on the 95% confidence level), which is much lower than recommended by the ICRU or stipulated in Hungarian National Standards. This accuracy is however sufficient in non-tumor therapy. The output variation among therapeutic X-ray units is high (sigma x approximately 22%) and correlates with the HVL measurements. The advantage of a central dosimetric service is the reduction of the number of unnecessary radiation qualities and superfluous combinations of radiation qualities, tube current and FSD.