A correlated sampling variance reduction technique has been implemente
d in the EGS4 (Electron Gamma Shower version 4) Monte Carlo code syste
m in order to calculate correction factors for radiation dosimeters. T
hese correction factors are calculated as the ratios of absorbed doses
in various geometries irradiated by photon and electron beams. In thi
s paper the efficiency and accuracy of the calculation of the absorbed
dose ratios with correlated Monte Carlo sampling is discussed. Theore
tical analysis shows that the calculation efficiency and accuracy incr
eases with the degree of correlation between the calculated doses in t
he individual geometries. Calculations of the absorbed doses in thin s
lab geometries and in air-filled ionization chambers have been perform
ed. The results show that the gain in computing efficiency is up to a
thousand fold for the calculations of the effect of the central electr
ode material and size on the response of cylindrical ionization chambe
rs. The calculation accuracy (systematic uncertainty) is significantly
improved when good correlation exists between the calculated doses in
similar thin-slab geometries.