This feasibility study was performed to evaluate the suitability of MRI in defining appropriate pelvic radiotherapy treatment volumes, and to compare MRI sequences with CT for prostate cancer radiotherapy. Five patients with localized prostate cancer, imaged with four MRI sequences (spin echo (SE) T1, turbo SE (TSE) T2, high resolution TSE (HR) T2, and FLASH 3D (F3D)), compared with their corresponding CT planning scans. Segmentation ability of the following pelvic structures: prostatic apex (PA), prostate, rectum, bladder and seminal vesicles (SV), were evaluated by three independent observers. They used a five point grading scale based on the anatomical definition of the organ boundary, tissue contrast and multiplanar display. Results were averaged for the group and for each sequence. There was no significant interobserver variation in the assessed scores (p > 0.1). The average scores (+/- 1 SD) for all pelvic structures assessed by each imaging sequence were CT 1.3 +/- 0.6; SE T1 2.4 +/- 0.9; TSE T2 2.4 +/- 0.7; HR T2 2.2 +/- 0.7 and F3D 3.4 +/- 0.6. Compared with CT, the average MR score for each assessed pelvic structure was higher with a trend for all transaxial MR sequences to provide improved segmentation of the PA and rectum. The F3D sequence scored highest as it provided multiplanar views and avoided the problem of partial volume averaging. MRI, compared with CT, appears to provide improved definition of pelvic treatment volumes but further work is required to confirm this and to address the issues of MRI associated distortion and dosimetry before MRI can be used routinely for pelvic radiotherapy planning.