California Bearing Ratio (CBR) is an important parameter used to evaluate the strength of subgrade and sub-base soils for design of flexible pavements and hence it plays a significant role in road and highway constructions. Obtaining CBR is heavily time consuming and it is difficult to acquire a representative CBR value. Therefore, many correlations have been developed by various researchers worldwide to predict the CBR. Due to differences in soil formations in the tropical environment, these existing global correlations found to be not satisfactory with local soils in Sri Lanka. Hence, this study was carried out to develop empirical correlations between CBR and index properties those best suit for local soils, using the data obtained from Atterberg limits and sieve analysis tests together with compaction tests. The new correlations were established using the method of regression analysis in the form of empirical equations representing the role of index properties. Robust regression by the method of least absolute residuals using MATLAB was considered in the analysis to reduce the impact of outliers along with traditional multiple regression using Microsoft Excel. As a final verification, several laboratory tests were conducted to compare the results with proposed regression equations.