Missing data are prevalent in criminological studies, especially in the large-scale surveys. However, in the field of criminology, little attention has been paid to the missing data issues. This study aims to provide a comprehensive evaluation of missing data problems and practical guidelines on how to choose the appropriate method to handle the missing data in criminology. The basic concepts and taxonomies in missing data problems are reviewed, such as types of nonresponse, patterns of nonresponse and missing data mechanism. The author also reviews the current literature on the strategies for handling missing data in the special context of criminological research and summarizes the advantages and disadvantages of each method. A simulation study is designed to evaluate the performance of three different methods: mean substitution, maximum likelihood with expectation maximization algorithm and multiple imputation. The results indicate that when the proportion of missing data is lower than .05 and data are MCAR, mean substitution is acceptable to handle missing data, while maximum likelihood with expectation maximization algorithm and multiple imputation are recommended, when the proportion of missing data is higher than .05. The author suggests that besides reporting how missing data problem is handled in criminological studies, more attention should be paid to the differences due to the choice of way of handling missing data.