The portal films of 54 consecutive patients treated for primary prosta
te cancer have been compared to the simulation films. The systematic a
nd random uncertainty in the set-up, defined by the couch movement req
uired to move the patient to the simulated position, was determined to
be 1.6 mm UP (SD 3.3 mm), 0.3 mm RT (SD 2.6 mm) and 1.3 mm IN (SD 2.4
mm). The area of fields defined on simulation films was compared with
that on portal films to determine the error in block production which
was -0.7 mm (SD = 0.9 mm). Five sources of uncertainty in the radioth
erapy have been identified, three occur before and two during the cour
se of treatment. A method for combining these uncertainties is propose
d and used on the data obtained in this study. This provides estimates
of the margin required when drawing blocks so that the minimum dose t
o the target is 95% of the prescription in 95% of treatments. The bloc
k margins are not uniform and range from 21 mm, when drawing the block
outline to the posterior on a lateral film, to 13 mm when drawing lat
erally on an anterior film.