Purpose: At the Wellington Blood and Cancer Centre (WBCC), Volumetric Modulated Arc Therapy (VMAT) is used to treat a variety of head and neck (H&N) cancers. Presently, the complexity of plans is limited to ensure the accuracy of patient treatment within the range of the departmental experience. The complexity limitation is applied through use of a monitor unit (MU) constraint during plan optimisation. Plans of higher complexity can be obtained by loosening the MU constraint, and setting more stringent optimisation objectives on organs at risk (OAR) and target volumes (PTV). This could potentially yield higher quality treatment plans but may also degrade the accuracy of the TPS calculation or the plan delivery at the treatment machine. The aim of this study is to investigate the level of plan complexity that results in accurate treatment plan calculation and delivery, and quantify the corresponding gain in plan quality. Methods: Five previously treated H&N patients were selected for the study. Each patient’s clinical plan was used as the lowest complexity level and labelled C1. Subsequently, an approximate pareto-optimal plan (C3) was created that focused equally on sparing spinal cord, brain stem and parotid gland while maintaining, or improving on, the previously obtained target coverage. Next, a C2 plan was created such that the plan quality was in between C1 and C3. Plan quality of each complexity level was assessed in terms of OAR sparing and PTV coverage. The average leaf pair opening (LPO), critical leaf pair opening (%LPO<1cm) and mean leaf travel were used as plan complexity metrics. The calculation and delivery accuracy of each complexity level using Varian TrueBeam LINAC/Eclipse TPS was verified using time resolved point dose measurements (TRPD), EBT film measurements (Ashland Inc.) and ArcCheck measurements (Sun Nuclear Corp.). A comprehensive uncertainty analysis was carried out including a quantification of the measurement and delivery reproducibility. Results: Increasing plan complexity from C1 to C3 reduced the Spinal Cord D1cc, Brain Stem D1 and Parotid Gland Dmean up to 14.7 Gy, 7.1 Gy and 7.8 Gy, respectively. In addition, C3 plans improved the target coverage compared to C1 plans, with the PTV66 and PTV54 D98 increasing up to 1.0 Gy and 0.6 Gy, respectively. The verification measurements showed that the plan calculation and delivery for all complexity levels was well within clinical acceptance levels (Table 1). TRPD showed that VMAT dose delivery itself was repeatable within 0.1% (1 S.D.) over 10 consecutive deliveries for both C1 and C3 complexity levels. Discussion & Conclusions: This study has shown that increasing the plan complexity can provide significant dosimetric advantages for the treatment of H&N cancer. Verification measurement results indicated that this did not noticeably degrade the calculation and delivery accuracy of VMAT using a Varian TrueBeam LINAC and our Eclipse TPS beam model. H&N VMAT at the WBCC can now be developed further with greater confidence in the dosimetric accuracy of higher complexity plans.