Lung function depends on mechanical lung expansion and contraction during the respiratory cycle. Recently developed dynamic 4D CT imaging and 3D image registration can be used to analyze regional lung function, which are significant for lung disease diagnosis, treatment and lung ventilation change during radiation therapy. 4D CT images of the lung can be reconstructed at any respiratory phase point based on the breathing trace signal during image acquisition. In this paper, we propose an image registration-based technique for assessing regional pulmonary function during specified time interval and estimating regional expiration volume ratio using 4D CT. Data from four anesthetized mechanically-ventilated sheep and one human patient undergoing Radiation Therapy were analyzed. Sheep lung images were divided into 30 slabs in the ventral-dorsal direction with equal lung height to study the ventilation variation. We found that the mean values of expiration volume ratio across slabs demonstrate a similar pattern and lung tissue near dorsal part contracts more in the first expiratory phase. The reproducibility of regional lung function of one human patient was compared between end-expiration-to-end-inspiration pair and the phase pair spanning the first expiratory interval, in which the latter pair shows 30% higher reproducibility. The accuracy of image registration is assessed by 200 semi-automatically annotated lung landmarks. Sheep data show landmark error on the level of 0.8 mm after registration.