Research on survey response errors has generally been based on data from household surveys. While establishment surveys are subject to many of the same errors, the two types of surveys differ in the type of data requested and the respondent role in providing those data, leading to errors that are unique to establishment surveys. In this paper, we identify and evaluate response errors in mail surveys of establishments, using data from five Response Analysis Surveys (RAS) conducted by the U.S. Bureau of Labor Statistics. The RAS studies assess the availability of data in the establishment's information system, compatibility of those data with survey definitions, respondent understanding of the questions and instructions, and respondent adherence to survey definitions in providing data. While most of the errors identified relate to respondent comprehension, others are directly associated with the content or structure of data in the information system.