The paper introduces the implementation of Multiple Imputation by Chained Equations (MICE) in Stata, a statistical software. MICE is a method for handling missing data by creating multiple imputed datasets, each generated using a different imputation model. The authors, Patrick Royston and Ian R. White, describe the MICE algorithm and its application in Stata through the `ice` command. They provide a detailed example using real data from an observational study on ovarian cancer to illustrate various options available with `ice`. The paper covers the structure of the `ice` system, including the `ice`, `ice_`, and `uvis` ado-files, and discusses the imputation models for different types of variables. It also explains how to fit analysis models using the `mim` command and combines estimates across imputed datasets using Rubin's rules. The paper highlights the flexibility and wide range of options provided by `ice`, noting its limitations and ongoing improvements. Additionally, it touches on new features in Stata versions 11 and 12, such as the MI system and the ability to handle monotone missing data patterns.The paper introduces the implementation of Multiple Imputation by Chained Equations (MICE) in Stata, a statistical software. MICE is a method for handling missing data by creating multiple imputed datasets, each generated using a different imputation model. The authors, Patrick Royston and Ian R. White, describe the MICE algorithm and its application in Stata through the `ice` command. They provide a detailed example using real data from an observational study on ovarian cancer to illustrate various options available with `ice`. The paper covers the structure of the `ice` system, including the `ice`, `ice_`, and `uvis` ado-files, and discusses the imputation models for different types of variables. It also explains how to fit analysis models using the `mim` command and combines estimates across imputed datasets using Rubin's rules. The paper highlights the flexibility and wide range of options provided by `ice`, noting its limitations and ongoing improvements. Additionally, it touches on new features in Stata versions 11 and 12, such as the MI system and the ability to handle monotone missing data patterns.