The Bureau of Labor Statistics (BLS) measures monthly job fluctuations by tabulating employee counts supplied by establishments on two different reports: State Unemployment Insurance Quarterly Contribution Reports and the BLS monthly Report on Employment, Payrolls, and Hours. For statistical purposes, BLS defines an employee as a person "who worked during or received pay for the pay period containing the 12th of the month." Participants often obtain these employee counts from reports associated with payroll records. Since many payrolls are generated by computer software, the definition of employment used by the software in producing counts has implications for the quality and accuracy of this vital economic indicator. In this paper, we report research into commercial payroll software produced by two types of organizations: payroll processing firms (service bureaus) that use proprietary software to generate paychecks and other payroll data, and software developers that sell or distribute payroll software to individual users. Software users can be private companies, or they can themselves be service bureaus or accountants who use the software to prepare paychecks for clients. The research focuses on measurement error associated with commercial payroll software and whether employee counts meet the method, timing, and content components of the employment definition.