To assess the variability of pollen emission, ten field experiments were carried out in July and August over four years using three crop varieties. The model was built from (i) a parameterisation of the measured diurnal and seasonal patterns of pollen emission followed by (ii) a quantification of the relationships between parameter values and meteorological conditions. Total production of pollen was fairly constant for a given variety over years and sowing dates, while patterns of emission varied with meteorological conditions. The pollen emission season was longer when temperature was low and humidity was high. In most cases, the diurnal pattern was unimodal: the onset and the peak of emission were delayed when relative humidity was high or wind speed was low. In some cases, a second peak of emission occurred during the afternoon. This was observed more frequently when temperatures were high.