Accurate prediction of construction cost in the initial phase of a construction project is critical to the success of the project. Accordingly, many researchers have proposed various methodologies for predicting the cost in the initial phase with the use of limited information. This study was aimed at improving the prediction performance of a cost prediction model based on the Case-Based Reasoning (CBR) technique, which has recently become widely used. Toward this end, an improved CBR model that uses the Multiple Regression Analysis (MRA) technique in the revision phase of the CBR technique was developed. To verify the prediction performance of the proposed model, a case study was performed on 41 business facilities and 99 multi-family housing projects. The results showed that the prediction performance of the revised CBR model for business facilities and multi-family housings improved by 17.23% and 4.39%, respectively, compared to that of the existing CBR model. The proposed MRA-based revised CBR model is expected to be useful in estimating the construction cost in the initial phase of a projec