VMAT is an important tool in the treatment of head and neck cancers, many of which also require treatment to the supraclavicular lymph nodes. However, full VMAT arcs treating this nodal region necessarily cause entrance beam to pass through patients' shoulders. Thus, interfractional variations in shoulder position may cause unwanted dose perturbations. To assess this possibility, six patients undergoing treatment at our institution for head and neck cancers with associated supraclavicular lymph node treatment were imaged with in‐room CT‐on‐rails during the course of their treatments. This allowed for the establishment of a true record of the actual shoulder position during selected treatment fractions. Then, a full VMAT plan and a plan with VMAT arcs superior to the shoulder and a static anteroposterior field inferiorly were copied onto the patients' weekly image sets. The average one‐dimensional shoulder motion was generally within 10 mm of the simulated position, with some notable exceptions. The standard deviation in week‐to‐week shoulder position relative to simulation was 4.3 mm and 4.2 mm in the SI and AP dimensions, respectively. The average nodal target mean dose across all fractions sampled was within 5% of planned for all patients and both plans. Similarly, the average D95 for the nodal target was within 5% of planned across all fractions sampled, with the single exception of the full VMAT plan for one patient. In most cases, the standard deviation in both target mean dose and D95 was smaller with the VMAT+static AP field plan than it was with the full VMAT plan.PACS number: 87.55.‐x