This study aims to develop models that may be used to describe the relationship between meteorological variables and ambient concentrations of pollen. We used daily ambient concentrations of grass pollen during the pollen season (October to December) of 2004 in Melbourne, Australia. During this period, daily levels of meteorological data including average relative humidity, mean temperature, rain fall, wind speed, and wind direction were entered as predictors in the models. A generalized additive model (GAM) was used to assess the relationship between daily levels of meteorological variables and ambient concentrations of grass pollen. The relationship between average temperature, rain fall, wind speed, relative humidity and pollen were nonlinear and smooth terms were highly significant (p < 0.001). Nonlinear statistical methods such as the GAM approach have the potential to accurately predict ambient concentrations of pollen during the pollen season.