Trailing edge serrations are a widely used passive technique for the suppression of aerodynamic noise from wind turbines. Despite their popularity, no reliable engineering prediction tool has yet been developed to estimate the noise reduction for different serrations. This paper concerns the development of an engineering noise prediction tool, based on a recently developed mathematical model. Results show that the new model has several advantages over Howe's model, as it can take both destructive and constructive sound interference effects into account. Two surface pressure wavenumber-frequency models are implemented, namely Chase and TNO models, to demonstrate the sensitivity of the model to boundary layer characteristics. The boundary layer parameters needed in the wavenumber-frequency models are obtained using RANS CFD simulations. Far-field noise comparisons are provided between the proposed prediction tool and experimental data for a NACA0018 airfoil. A parametric study regarding the boundary layer changes of serrated airfoils signifies the need for more reliable wavenumber-frequency models. The results presented in the paper show that the proposed engineering tool can provide a fairly accurate estimate of the noise reduction performance of serrated airfoils, but its accuracy relies heavily on the availability of reliable near-field boundary layer information. (C) 2019 Elsevier Ltd. All rights reserved.