Existing parametric correlations have been found to have difficulties in predicting the removal of trace levels of volatile organic chemicals (VOCs) by modern air stripping towers. In this study a new approach using nonparametric kernel regression (NKR) method was used to predict the mass transfer coefficient of air stripping towers. Though, only four variables were used, the predictions are already improved by more than 50% as compared to Onda correlation, the best existing parametric correlation. The proposed technique shows a dependency of KLa on the liquid flow rate which is in good agreement with established theory. Previous parametric approaches were unable to model correctly this relationship.