The aim of this study was to assess portal imaging for quality assurance of patient positioning in external beam radiotherapy. We present a retrospective study of the variability of patient position in the treatment of 34 prostate cancer patients who were treated with whole pelvic irradiation followed by arc therapy or boost field (Series I) and 25 patients treated by 'small' pelvic 4-field box technique (Series II). Weekly anteroposterior-posteranterior (AP-PA) and left-lateral portal images were compared to simulation films by using a fiducial point-pair registration technique based on the computer-assisted portal imaging quality assurance program PIPSpro, developed specifically for the verification of treatment positioning in radiation therapy. Series I consisted of 34 patients and 194 portal films (97 AP-PA and 97 left-lateral). Overirradiated (OA) and underirradiated (UA) areas were computed in terms of percentage of the reference field size. For the AP-PA portals, the average OA was 2.75% and average UA was 2.74%. For left-lateral portals, an average OA of 2.49% and UA of 2.78% were measured. Series II consisted of 25 patients and 194 portal films (98 AP-PA and 96 left-lateral). The average OA was 0.88% and average UA was 0.86% in AP-PA portals, and 1.03 and 0.82% for left-lateral portals, respectively. The accuracy of patient positioning in irradiation of prostate cancer in our institution is in the range of 2.69% for whole pelvic fields and 1.0% for small fields. We conclude that PIPSpro is an effective and useful tool for quality assurance in radiotherapy.