Aboveground biomass (AGB) is a vital factor when evaluating the grassland ecosystem service values, especially on regional scales. It is often estimated using the Carnegie-Ames-Stanford Approach (CASA). However, the significant spatial heterogeneity of landforms in karst areas poses limitations for CASA, resulting in low accuracy in estimating AGB. The objective of this study was to combine CASA model with karst bare rock index (KBRI) to improve the estimation accuracy of grassland AGB in a typical karst area, Southwest China. The coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE) and modelling efficiency (ME) of the improved CASA are 0.677, 76.735 g m-2, 65.224 g m-2 and 0.736, respectively. These values are higher than those of the original CASA model, which has an R2 of 0.468, RMSE of 98.541 g m-2, MAE of 71.434 g m-2, and ME of 0.519. The improved CASA model showed an average value of 739.95 g m-2, which is closer to the measured AGB. Therefore, it is feasible to estimate the grassland AGB in karst area by integrating karst bare rock index into CASA model. Our results have provided new insights for monitoring grassland productivity in karst areas.