Abstract We present an uncertainty measure that is based on a business survey in which uncertainty is captured directly by a qualitative question on subjective uncertainty regarding expectations. Uncertainty perceptions display persistence at the firm level and changes are associated with past business assessments and expectations. While our uncertainty measure correlates with commonly used alternatives, it is superior in forecasting and suggests a larger role of uncertainty shocks for aggregate fluctuations. Its informational content is highest when considering smaller firms or firms with a low growth rate. Our results confirm the feasibility of constructing uncertainty measures from business survey questions that elicit information on uncertainty of respondents directly.