A linear model has been used for the regression and prediction of daily air pollution levels con- structing an operational peak ozone concentra- tion forecasting module, developed from scratch, for the Athens basin, Greece. Maximum tempera- ture, maximum ozone level of the previous day and wind speed proved to be the most essential factors that influence ozone concentrations. Regarding the persistence model, it seems to be acceptable, but the extremely large proportion of false alarms should be considered together with the operational requirements of the system as the principal limitations.