This paper proposes a novel method to implement a first-order equivalent thermal parameter (ETP) model, a commonly used model to provide building thermal response prediction for advanced building controls. This method reformulates the existing implementation method to allow the E T P model to capture varying heat disturbance and non-linearity in the zone temperature evolution. It also guarantees that the prediction from the E T P model is always reasonable by considering more physical constraints. To evaluate performance, we implemented an E T P model w i t h the proposed method for a medium-office building and predicted the zone temperatures for one day. Prediction results are compared against those from an E T P model implemented through a common approach and a lookup-table model. Results suggest that the E T P model implemented by the proposed method can yield more accurate results compared to the other two models; additionally, the E T P model implemented by the proposed method demonstrates similar capabilities in capturing non-linear behaviors as the lookup-table model.