Purpose: To evaluate two deformable image registration (DIR) algorithms for the purpose of contour mapping to support image guided adaptive radiotherapy (IGART) with 4D cone beam CT (4DCBCT). Methods: Eleven locally advanced non‐small cell lung cancer (NSCLC) patients underwent one planning 4D fan‐ beam CT (4DFBCT) and seven weekly 4DCBCT scans. Gross tumor volume (GTV) and carina were delineated by a physician in all 4D images. For day to day registration, the end of inspiration 4DFBCT phase was deformably registered to the corresponding phase in each 4DCBCT image. For phase to phase registration, the end of inspiration phase from each 4D image was registered to end of expiration phase. The delineated contours were warped using the resulting transforms and compared to the manual contours through Dice similarity coefficient (DSC), false positive and false negative indices, and, for carina, target registration error (TRE). Two DIR algorithms were tested: 1) small deformation, inverse consistent linear elastic (SICLE) algorithm and 2) Insight Toolkit diffeomorphic demons (DEMONS). Results: For day to day registrations, the mean DSC was 0.59 ± 0.16 after rigid registration, 0.72 ± 0.13 with SICLE and to 0.66 ± 0.18 with DEMONS. SICLE and DEMONS reduced TRE to 4.1 ± 2.1 mm and 5.8 ± 3.7 mm respectively, from 6.2 ± 3.5 mm; and reduced false positive index to 0.27 and 0.26 respectively from 0.46. Registration with the cone beam as the fixed image resulted in higher DSC than with the fan beam as fixed (p < 0.001). SICLE and DEMONS increased the DSC on average by 10.0% and 8.0% and reduced TRE by 2.8 mm and 2.9 mm respectively for phase to phase DIR. Conclusions: DIR achieved more congruent mapping of target structures to delineations than rigid registration alone, although DIR performance varied with algorithm and patient. This work was supported by National Cancer Institute Grant No. P01 CA 116602.