Bureau of Labor Statistics' (BLS) staff used cognitive testing to evaluate and refine the questionnaire for the Current Employment Statistics (CES) Business Births Pilot Study. The CES survey is the source of current monthly U.S. payroll employment data. At present, it does not directly measure employment from newly-created businesses, or "business births." The means of selection into the CES sample is through an employer's Unemployment Insurance (UI) account number, and there is a lag of as much as 8 months between the time an employer obtains the account and the time it appears in the sampling frame. To collect data from business births near the time of their inception, BLS developed a procedure to sample businesses shortly after they obtain new UI accounts. The Business Births Pilot Study, getting underway in 1996, is a telephone survey designed to test the feasibility of differentiating new businesses from ongoing businesses that are undergoing changes such as new ownership, incorporation, or merger. The telephone interview will determine whether a business is a birth, and if so, will obtain employment data and an industry description for it. The paper reports on the types of questionnaire problems identified through cognitive interviews, and on serendipitous findings about respondents' understanding of basic survey concepts.