Résumé : In order to test the feasibility of a protocol for intra-fractional adjustment of the patient position, during radiation therapy treatment in the pelvic region, a two-fold study is carried out. The protocol involves an objective quantitative measurement of the error in positioning starting from the comparison of a portal image with a reference image. The first part of the study applies the protocol to determine the efficacy of adjustment using subjective determination of the positioning errors by a clinician by measuring the residual errors after adjustment. A group of 13 patients was followed extensively throughout their treatment, analyzing 240 fields. In the second part the measurement itself determines the extent of readjustment of the position. Throughout the procedure elapsed time is measured to determine the extra time involved in using this procedure. For this part a group of 21 patients was followed yielding statistics on 218 fields. Using this computer aided protocol it is shown that systematic as well as random errors can be reduced to standard deviations of the order of 1 mm. The price to pay however is additional treatment time up to 58% of the treatment time without the protocol. Time analysis shows that the largest part of the added time is spent on the readjustment of the patients' position adding a mean of 37% of time to the treatment of one field. This is despite the fact that the readjustment was performed using a remote couch controller. Finally a statistical analysis shows that it is possible to select patients benefiting from the use of such a protocol after a limited number of fractions. © 1995.