Abstract. In recent years, environmental concerns have encouraged the use of wind power as a renewable energy resource. However, high penetration of the wind power in the electricity system is a challenge due to the uncertainty of wind energy forecast. Estimation of the wind energy production requires a forecast for the wind (the main source of uncertainty) but also of density, often overlooked. Measure of air density is a key for more accurate wind energy prediction. Wind farms often lack instrumentations of temperature and pressure, needed for accurate air density estimation at hub height to be used for locally debiasing air density forecast. In this study, the error budget of air density estimate is computed distinguishing temperature and pressure contributions. The analysis uses measurements for in-depth local analysis as well as meteorological reanalysis to investigate the added-value of a model-based value when measurement is missing. Meteorological reanalysis is also used to study spatial pattern of error budgets (mountainous area, coastal regions, plains, ...). The effect of altitude is carefully accounted for. Temperature is by far the variable inducing the largest errors when it is missing in the air density correction, and replaced by the standard atmosphere value (i.e. 15 °C, used as reference in power curves). It is particularly true for very cold or warm conditions (i.e. far from the standard value), for which the error on wind energy production is nearly halved when an accurate correction of temperature is performed.