Nowadays in radiotherapy, much effort is taken to minimize the irradiated volume and consequently minimize doses to healthy tissues. In our work, we tested the hypothesis that the mean dose distribution calculated from a few fi rst fractions can serve as prediction of the cumulated dose distribution, representing the whole treatment. We made our tests for 25 prostate cancer patients treated with three orthogonal fi elds technique. We did a comparison of dose distribution calculated as a sum of dose distribution from each fraction with a dose distribution calculated with isocenter shifted for a mean setup error from a few fi rst fractions. The cumulative dose distribution and predicted dose distributions are similar in terms of gamma (3 mm 3%) analysis, under condition that we know setup error from seven fi rst fractions. We showed that the dose distribution calculated for the original plan with the isocenter shifted to the point, defi ned as the original isocenter corrected of the mean setup error estimated from the fi rst seven fractions supports our hypothesis, i.e. can serve as a prediction for cumulative dose distribution.