This study focuses on ozone modeling using meteorological and air monitoring variables. Twenty seven (27) places in Seoul were measured for ozone values from January 1999 to December 1999. Air quality monitoring data consisted of CO, NO2, O3, PM10, TSP while meteorology data consisted of the daily maximum temperature, humidity and wind speed, and solar radiation. The complexity of environmental data dynamics often requires models covering non-linearity. Photochemical ozone pollution is the result of complex non-linear interactions between atmospheric pollutants and meteorology. The generalized additive model is favored because it is the most flexible, has the fewest statistical assumptions, and it can detect and fit potentially complex and nonlinear dependencies. For these reasons we modeled the daily ozone amount using a generalized additive model with smooth loess functions and compared it with a multiple linear regression model.