Background: In cancers of the head and neck, gross tumor or areas at risk of microscopic disease often lie close to the skin, while the skin itself may not be at risk. With intensity-modulated radiotherapy, setup errors can lead to underdosage of superficial structures because the collimator will not by default open beyond the skin surface to apply coverage in the air overlying the skin. Thus, small setup errors can move superficial structures out of field for some beams. Some planning systems allow for manually extrapolating fluence for beams tangential to superficial targets. It is unclear whether this problem is significant with tomotherapy. Methods: A head and neck phantom was utilized. A 3-mm bolus was used to represent the skin and allow placement of dosimeters at 3 mm depth. Thermoluminescent dosimeters were placed at reproducible points on the skin surface and at 3 mm depth. The phantom was irradiated, with the target volume deep to the thermoluminescent dosimeters receiving a dose of 5 Gy. This process was repeated with the phantom displaced 2.5 mm and again with a displacement of 5 mm. These displacements simulated setup errors that in clinical practice would correspond to bending or twisting of the neck that could not be corrected with rotations or translations. Results: When the phantom was displaced 2.5 mm, the dose measured at 3 mm depth was 99.2% (95.9%-102.5%) of the control. With a 5-mm displacement, the dose at 3 mm only dropped to 91.1% (88.8%-93.4%) of the control. Dose measured at skin surface decreased to a greater degree with such setup error. Conclusions: Dose at superficial depths degraded only slightly with 2.5-mm and even 5-mm displacements. With the tomotherapy system, superficial dose appears to be robust to clinically relevant setup errors. However, if the skin is at risk, bolus should be used to ensure adequate coverage.