In pelvic radiotherapy, accurate patient positioning is essential to ensure the target area receives sufficient coverage and minimal radiation to the surrounding organs at risk. Although image-guided radiotherapy (IGRT) has significant advances, there are still some setup errors that happen because of patient positioning variability and their anatomical change between treatments, such as bladder filling or rectum stretches. Quantifying these uncertainties is essential for determining appropriate clinical target volume (CTV) to planning target volume (PTV) margins. This study aims to assess setup errors in pelvic cancer patients receiving IGRT and to determine suitable CTV–PTV margins through a population-based methodology. A retrospective study was conducted on 20 patients with pelvic malignancies, who received intensity-modulated radiotherapy or volumetric-modulated arc therapy, in a Halcyon IGRT-based treatment system with daily kilovoltage cone-beam computed tomography (KV-CBCT) imaging. Setup deviation was recorded and analysed in the lateral, longitudinal and vertical directions for each treatment fraction. Population mean errors, along with systematic (S) and random errors (s), were calculated, followed by the estimation of CTV–PTV margins using the Van Herk formula. The results showed minimal setup errors in the overall population mean. Systematic errors were relatively low across all directions, indicating good reproducibility of patient positioning, whereas random errors were comparatively greater, particularly in the longitudinal and lateral directions, indicating daily variation. The calculated CTV–PTV margins were 0.36, 0.78 and 0.59 cm in the vertical, lateral and longitudinal directions, respectively. The largest margin was observed in the lateral direction, indicating greater residual uncertainty despite daily image guidance. These findings indicate that although daily KV-CBCT-based IGRT effectively reduces setup uncertainties, residual variation still influences margin requirements. Institution-specific evaluation of setup errors remains important for optimising CTV–PTV margins and maintaining accurate target coverage while minimising radiation exposure to surrounding normal tissues.