Data analysis is an important link in the data governance of oil and gas exploration and development. Due to various problems in data collection, storage and processing, different departments and different specialties lack unified data management norms and standards, resulting in difficult to effectively connect data among departments and professional links and uneven data quality, which affects the accuracy and effectiveness of subsequent data analysis. Identify outliers to integrate, clean and analyze data, improve data reusability and traceability, reduce data management costs, and help improve the level of oil and gas exploration and development data governance. Based on multiple sets of data in CNOOC Data Lake, this study compares Grubbs method, Dixon method and robust statistical method in outlier analysis, and finds that robust statistical method can effectively identify outliers in oil and gas exploration and development data. In robust statistical methods, quartile method, iteration method and qualified data range determination method can effectively identify outliers. For individual cases that are difficult to judge, comprehensive consideration of the identification results of various methods is conducive to improving the reliability of outlier determination conclusions. Through the outlier analysis of various data generated in the process of oil and gas exploration, development and production, the efficient use and reasonable application of data can be realized, which is helpful to optimize the production plan, reduce the waste of resources in the production process, and improve the oil and gas production efficiency.