In this study we investigate the usefulness of business survey data in forecasting Hungarian manufacturing output growth in the short run. We analyse the individual questions of the business surveys, and use models with different flexibility (factor model, best fitting and recursively best fitting model) to estimate the relationship between the business survey indicators and manufacturing output growth. The models are evaluated according to their forecasting performance. We generally find that although confidence indicators can be useful in forecasting manufacturing output in the short run, their forecasting ability is limited to a one-quarter horizon. For this reason their use in forecasting should mainly be restricted to nowcasting purposes.