Objective Meta analysis was used to explore the relationship between chest CT features and clinical severity of 2019 novel coronavirus pneumonia, so as to provide basis for screening and diagnosis of 2019 novel coronavirus pneumonia patient. Methods "Embase" "PubMed" "Cochrane Library" was retrieved from December 27, 2019 to December 1, 2020, and the key words of 2019 novel coronavirus pneumonia were searched. The standard of the selected papers was in the same chest CT research literature, including non severe and severe patients. The data included chest CT lesion distribution, lesion composition, signs, mediastinum and pleural involvement. After data collection, Stata 12.0 software was used for data analysis. Results A total of 4105 people, 2639 in the non severe group and 1466 in the severe group were included in 23 articles. The statistical results showed that the CT features more likely occured in severe group were: consolidation, coexistence of ground glass opacity and consolidation, crazy-paving pattern, air bronchogram, bronchial wall thickening,traction bronchiectasis,parenchymal band,vascular enlargement, lymphadenopathy,bilateral distribution, central distribution, coexistence of peripheral and central distribution,right upper lobe, right middle lobe, right lower lobe and left lower lobe distribution. The non severe group was more likely to have nodules, while the severe group and the non severe group had no significant difference in CT signs, including ground glass opacity, interlobular septal thickening, reticular sign, peripheral distribution and left upper lobe distribution. Conclusion Lesion distribution, lesion composition and CT signs were different in non severe group and severe group of 2019 novel coronavirus pneumonia. In severe group, there were more consolidationlesions, ground glass opacity and consolidation mixed lesions, crazy-paving pattern, vascular enlargement, pleural effusion and so on.