The age of air is a useful integrated quantity that represents transport processes. Most atmospheric models underestimate the age of air, especially when wind fields from data assimilation systems are used. Nonetheless data assimilation is necessary to provide realistic winds. The European Centre for Medium-Range Weather Forecasts (ECMWF) uses different assimilation procedures with various forecast periods, where forecasting allows the model to recover from the non-physical adjustments of the assimilation procedure. In this study we examine the impact of different assimilation procedures and forecast periods on ECWMF wind fields by performing simulations of the age of air. For all data sets the age of air is too young, and for the 45-years re-analysis data (ERA-40) it is even anomalously young, which has to be attributed to the data assimilation procedure used in ERA-40. Extending the forecast period partially remedies the low ages of air, but to an unsatisfactory degree.