Large gas fields play an important role in natural gas industry. Recovery rate, plateau duration, recovery at the end of plateau, decline rate and recovery factor are the key development indexes for dynamic performance analysis and development planning. Scientific prediction for those indexes can support gas development planning strongly. Through mining statistical analysis of 150 large gas fields and numerical simulation analysis, 23 objective influencing factors which affect the development effect are studied. Gas fields are classified according to the main influencing factors, and then the distribution of development indexes are summarized. Finally, a series of prediction methods for key development indexes are established. Based on the above work, it is found that matrix permeability, drive types, reservoir architecture and fluid type are the most sensitive factors among them. According to the most sensitive factors, gas fields should be divided into 4 categories, and 13 subcategories and the distributions rules of development indexes of all categories are presented. Then new prediction methods for development indexes are established, including linear empirical formula method, similarity analogy prediction method based on Euclidean theorems, and probabilistic values method. In this process, according to the characteristics of influencing factors, logarithmic and piecewise function methods are used for dimensionless treatment, and the prediction accuracy of the methods is improved. Finally, the expert system software is developed which can automatically predict the key development indexes. The prediction accuracy is over 80% which can satisfy the requirement of strategic planning. The new methods have the characteristics of multiple methods, applicable to multiple gas field types and predicting multiple development indexes. Those methods can be applied to predict the development indexes of new fields and evaluate the development effects of matured gas fields in batch.