Purpose: Multiple anatomy optimization (MAO) utilizing deformable dose accumulation on entire 4DCT data sets is implemented to overcome ambiguity between optimal dose defined on a single anatomy and optimal accumulated dose resulting from dose delivery to moving and deforming anatomy. Methods: Six lung cancer patients are planned using two methods of radiotherapy optimization: the internal target volume (ITV) envelope method and MAO, which simultaneously optimizes a single fluence for delivery to all 10 breathing phases such that the accumulated dose satisfies the plan objectives. Target dose is constrained to 70 Gy. The ITV‐plan is optimized on a single breathing phase with the planning target volume defined as the ITV; the MAO target is the moving CTV. MAO is compared to single image ITV optimization based on the accumulated dose assuming equal monitor‐units to each phase. Dose‐volume differences between single image estimations and 10‐image accumulation are examined. Results: Single image optimal dose distributions overestimate target V70 by 4.2%±3.1% (average, one standard deviation) and in five of six cases ipsilateral lung V20 is underestimated (1.4%±0.9%). For these five cases, MAO increases V70 by 2.8%±2.5% (maximum of 6% increase in V70) and reduces ipsilateral lung V20 by up to 3% (average decrease of 1.2%±1.3%). Contralateral lung V20, esophagus V25, and heart V30 are also reduced by up to 5%, 3%, and 3%. For the sixth case, lung tumor motion is on the order of the dose voxel size (3mm), and MAO did not improve upon the ITV plan. Conclusions: Dose‐volume optimization on a stationary image does not ensure accumulated dose coverage to the moving CTV. Multiple anatomy optimization can remove dose ambiguity and improve plan quality. P01CA11602 and Philips Medical Systems.