This study compares the evidence on corruption between alternative data sets. These include the Corruption Perceptions Indices (CPI) that are conventionally used and the micro data sets from the International Crime Victim Surveys (ICVS) and the World Bank Enterprise Surveys (WBES) that have been used in recent applications. While a comparison between the evidence from the CPI and WBES constitutes a comparison of perception versus reality, the comparison of evidence from ICVS and WBES can be construed as a comparison of individual with business corruption. The study finds several similarities and differences between the pictures on corruption yielded by the alternative data sets. For example, while in case of low income countries, perception of business corruption seems to be worse than that based on firms’ actual experience of doing business there, the reverse is true for high income countries. The magnitude of individual corruption is consistently lower than that of business corruption, with the gap between the two forms of corruption closing only for high income countries. As a country develops and commercial transactions increase, the mix of corruption changes in favour of business corruption. While the study finds evidence of a negative association between per capita GNP and corruption rates, none of the three data sets provides any evidence of negative association between growth and corruption rates. The study also finds that while improvement in human development indicators such as literacy are effective instruments in controlling individual corruption, the strengthening of institutions such as the legal system and the regulatory mechanism are likely to be more effective in combating business corruption. The strengthening of trust, whether via improved literacy and development of social networks or via a strong legal system, and an effective and transparent regulatory mechanism is the key to combating both forms of corruption. A methodological contribution of this study is the combination of the information of the characteristics of the respondent with the country level indicators in analysing the determinants of corruption. A significant difference between the two forms of corruption is that, after controlling for the respondent’s attributes and the country indicators, while individual corruption showed an increase over time, this was not the case with business corruption. The importance of introducing the country effects is seen from the sign reversal of the time coefficient estimate that occurs in case of both individual and business corruption once we control for the effects of the country of residence of the respondent. The overall message of this study is that the authorities need to distinguish between different forms of corruption in devising policy intervention. As the mix of individual and business corruption changes with economic development, so should the mix of policy instruments in tackling corruption. The results also underline the need to undertake more studies that investigate the sensitivity of the evidence on corruption to alternative data sets.