Abstract Background and purpose: Body contours can potentially be used for patient set-up verification in external-beam radiotherapy and might enable more accurate set-up of patients prior to irradiation. The aim of this study is to test the feasibility of patient set-up verification using a body contour scanner. Material and methods: Body contour scans of 33 lung cancer and 21 head-and-neck cancer patients were acquired on a simulator. We assume that this dataset is representative for the patient set-up on an accelerator. Shortly before acquisition of the body contour scan, a pair of orthogonal simulator images was taken as a reference. Both the body contour scan and the simulator images were matched in 3D to the planning computed tomography scan. Movement of skin with respect to bone was quantified based on an analysis of variance method. Results: Set-up errors determined with body-contours agreed reasonably well with those determined with simulator images. For the lung cancer patients, the average set-up errors (mm)±1 standard deviation (SD) for the left-right, cranio-caudal and anterior-posterior directions were 1.2±2.9, −0.8±5.0 and −2.3±3.1 using body contours, compared to −0.8±3.2, −1.0±4.1 and −1.2±2.4 using simulator images. For the head-and-neck cancer patients, the set-up errors were 0.5±1.8, 0.5±2.7 and −2.2±1.8 using body contours compared to −0.4±1.2, 0.1±2.1, −0.1±1.8 using simulator images. The SD of the set-up errors obtained from analysis of the body contours were not significantly different from those obtained from analysis of the simulator images. Movement of the skin with respect to bone (1 SD) was estimated at 2.3 mm for lung cancer patients and 1.7 mm for head-and-neck cancer patients. Conclusion: Measurement of patient set-up using a body-contouring device is possible. The accuracy, however, is limited by the movement of the skin with respect to the bone. In situations where the error in the patient set-up is relatively large, it is possible to reduce these errors using a computer-aided set-up technique based on contour information.