Much empirical work in the social sciences relies on the accuracy of survey responses. Of all the questions answered by survey respondents, few are as common as those concerning income: Income is a crucial determinant of an individual’s attitudes and behaviors and a standard correlate in political science survey research. This paper uses Danish administrative records to identify systematic error in survey respondents’ self-reported income. We show that income overreporting is most pronounced among individuals who share the characteristics of high-income individuals, in ways that suggest the presence of social desirability bias. Further, this leads to biased estimates and distorted conclusions in a number of common applications in political science, but a simple logarithmic transformation eliminates the bias. More broadly, our results indicate that to understand the income–attitudes nexus in a given context, survey researchers should take into account social desirability bias and the pattern of income misreporting in that context.