Abstract In mineral beneficiation, grinding is the final stage in the process of size reduction. The power consumed in this stage is higher when compared to other stages, owing to increased size reduction ratio. The primary purpose of grinding is to reduce the particle size to optimum so that mineral particles can be extracted more economically. Decision making plays an important role here, as it involves determining and comparing the energy that is required to perform the grinding process and also determining the amount of minerals lost as the coarser size particles are arrived at in mineral beneficiation. In general, Bond's work index is used to determine the grinding efficiency and also to calculate the power requirement. The process is very time consuming and it requires skilled labor and specialized mill. A systematic investigation was carried out to predict Bond's work index using simple field measurable properties of rocks. Tests were conducted on Basalt, Slate and Granite using a laboratory scale ball mill and rock properties namely density, Protodyakonov's strength index and rebound hardness number were determined. The results were analyzed using artificial neural networks and regression analysis. Mathematical equations were developed to predict Bond's work index based on rock properties using regression analysis, which resulted a very good correlation co-efficient values.