Purpose: Many techniques of delivering multiple isocentric IMRT or IMAT have been described. However, these studies have considered different treatment sites, reported different dosimetric parameters, and considered a limited degree of junction error, typically a 1D beam placement error of 3 mm. This work provides a direct comparison of these techniques by utilizing a consistent set of dosimetric parameters and a common treatment, craniospinal irradiation. Methods: Seven IMAT treatment plans were created, with three main features: (Method1) target subvolumes used to explicitly control the transition of dose from one set of beams to another in the junction region, (Method2) staggering the jaw settings for a set of arcs that overlap with another set of arc fields but not directly controlling the dose transition, (Method3) arranging beams to overlap 5–10 cm and allowing the inverse planning system to control the junction dose. Method3 was subdivided into three and four isocenter plans; four isocenter plans had greatest beam overlap. Then, the beam isocenters were off set 0.25 and 0.5 cm in each dimension (3D); therefore junction errors of 1.0 cm or greater were considered. For all plans, a total of 6 junction errors were considered. Results: Dosimetric impact of junction error was limited mostly to target volumes. The PTV V95 degraded by 6% to 9% and 1% to 3% for the 0.5 and 0.25 cm junction errors, respectively. For the CTV the V95 was impacted by 0.5% to 2.6% and 0.1% to 0.7%, respectively. The four isocenter plans of Method 3 consistently showed the largest degradation in V95 and V105 for both the PTV and CTV. Method1 and three field plans of Method 3 showed the least change in V95 and V105. Conclusion: Robust multiple isocenter IMAT plans, in terms of junction error, can be achieved with proper inverse planning techniques.