Abstract The delineation of target volume and organs at risk depends on the organs definition, and on the modalities for the CT-scan acquisition. Inter-observer variability in the delineation may be large, especially when patient’s anatomy is unusual. During the two french multicentric studies of conformal radiotherapy for localized prostate cancer, it was made an effort to harmonize the delineation of the target volumes and organs at risk. Two cases were proposed for delineation during two workshops. In the first case, the mean prostate volume was 46,5 mL (extreme : 31,7-61,3), the mean prostate and seminal vesicles volume was 74,7 mL (extreme : 59,6-80,3), the rectal and bladder walls varied respectively in proportion from 1 to 1,45 and from 1 to 1,16 ; in the second case, the mean prostate volume was 53,1 mL (extreme : 40,8-73,1), the volume of prostate plus seminal vesicles was 65,1 mL (extreme : 53,2-89), the rectal wall varied proportionally from 1 to 1, 24 and the vesical wall varied from 1 to 1,67. For participating centers to the french studies of dose escalation, a quality control of contours was performed to decrease the inter-observer variability. The ways to reduce the discrepancies of volumes delineation, between different observers, are discussed. A better quality of the CT images, use of urethral opacification, and consensual definition of clinical target volumes and organs at risk may contribute to that improvement.