This article presents estimates of six dimensions of governance for 199 countries and territories for 1996, 1998, 2000, and 2002. The data are based on 250 individual measures from 25 sources produced by 18 organizations. These measures are organized into six clusters corresponding to six dimensions of governance. An unobserved-components model is used to construct six aggregate governance indicators for each period. Point estimates and margins of error for each country are provided. Methodological issues, including potential biases and the interpretation of data, are also addressed. The data and a Web-based interface are available online. The six dimensions of governance include: voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law, and control of corruption. Each dimension is measured using a combination of subjective and objective data sources. The data are aggregated using an unobserved-components model, which allows for the computation of margins of error for each indicator. The model assumes that the only reason two sources might be correlated is that both are measuring the same underlying unobserved governance dimension. The governance indicators are constructed using data from a variety of sources, including polls of experts, surveys of businesspeople or citizens, and assessments of country performance by the World Bank. The data are aggregated to produce six aggregate indicators for each period. The indicators are based on the assumption that the distribution of governance in each country is normal, conditional on the data for that country. The conditional mean and standard deviation of the distribution are used to estimate the governance level for each country. The margins of error associated with the governance indicators are substantial relative to the units in which governance is measured. This means that comparisons across countries and over time should be made with caution. The margins of error are computed using the unobserved-components model, which allows for the computation of margins of error for each indicator. The model assumes that the only reason two sources might be correlated is that both are measuring the same underlying unobserved governance dimension. The article also discusses the use of perceptions-based sources and potential ideological biases in the data. It notes that subjective data are often used because objective data are difficult to obtain for many dimensions of governance. The article also discusses the importance of margins of error in the interpretation of governance indicators and the need for caution in interpreting changes in governance estimates over time. The article concludes that the governance indicators provide a useful tool for measuring governance, but their interpretation requires careful consideration of the margins of error and the potential for ideological bias.This article presents estimates of six dimensions of governance for 199 countries and territories for 1996, 1998, 2000, and 2002. The data are based on 250 individual measures from 25 sources produced by 18 organizations. These measures are organized into six clusters corresponding to six dimensions of governance. An unobserved-components model is used to construct six aggregate governance indicators for each period. Point estimates and margins of error for each country are provided. Methodological issues, including potential biases and the interpretation of data, are also addressed. The data and a Web-based interface are available online. The six dimensions of governance include: voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law, and control of corruption. Each dimension is measured using a combination of subjective and objective data sources. The data are aggregated using an unobserved-components model, which allows for the computation of margins of error for each indicator. The model assumes that the only reason two sources might be correlated is that both are measuring the same underlying unobserved governance dimension. The governance indicators are constructed using data from a variety of sources, including polls of experts, surveys of businesspeople or citizens, and assessments of country performance by the World Bank. The data are aggregated to produce six aggregate indicators for each period. The indicators are based on the assumption that the distribution of governance in each country is normal, conditional on the data for that country. The conditional mean and standard deviation of the distribution are used to estimate the governance level for each country. The margins of error associated with the governance indicators are substantial relative to the units in which governance is measured. This means that comparisons across countries and over time should be made with caution. The margins of error are computed using the unobserved-components model, which allows for the computation of margins of error for each indicator. The model assumes that the only reason two sources might be correlated is that both are measuring the same underlying unobserved governance dimension. The article also discusses the use of perceptions-based sources and potential ideological biases in the data. It notes that subjective data are often used because objective data are difficult to obtain for many dimensions of governance. The article also discusses the importance of margins of error in the interpretation of governance indicators and the need for caution in interpreting changes in governance estimates over time. The article concludes that the governance indicators provide a useful tool for measuring governance, but their interpretation requires careful consideration of the margins of error and the potential for ideological bias.