PURPOSE To evaluate the magnitude of systematic and random errors from a subset of 100 prostate and 26 head and neck (H&N) cancer patients treated with conventional conformal radiotherapy and using image-guided radiotherapy (IGRT). After treatment, the uncertainties involved and the CTV to PTV margin were evaluated. MATERIAL AND METHODS An Elekta Synergy庐 linear accelerator was used, taking advantage of 3D on-board computed tomography. IGRT with no-action level (NAL) protocol was applied, reporting the 3D translation and rotation corrections. A statistical study was performed to analyse systematic, random and interobserver uncertainties, and, finally, to obtain the CTV to PTV margins. RESULTS The H&N patients' uncertainties found were smaller than those of prostate patients. The CTV to PTV margins assessed, following the guidelines found in the literature, in the three dimensions of space (right-left, superior-inferior, anterior-posterior) were (5.3, 3.5, 3.2) mm for H&N and (7.3, 7.0, 9.0) mm for prostate cancer treatments. CONCLUSIONS It was found that assessing all the involved uncertainties within radiation treatments was very revealing; their quality improves using IGRT techniques and performing extensive data analysis.