This paper deals with the topic of revision of data with the aim of investigating whether consecutive releases of macroeconomic series published by statistical agencies contain useful information for economic analysis and forecasting. The rationality of the re-visions process is tested considering the complete history of data and an empirical application to show the usefulness of revisions for improving the precision of forecasting model is proposed. The results for Italian GDP growth show that embedding the revision process in a dynamic factor model helps to reduce the forecast error.