A critical process in brass casting is blending of the raw materials in a furnace so that the specified metal ratios are satisfied. The uncertainties in raw material compositions may cause violations of the specification limits and extra cost. In this study, we proposed a chance-constrained stochastic programming approach for blending problem in brass casting industry to handle the statistical variations in raw material compositions. The proposed approach is a non-linear mathematical model that is solved global optimally by using GAMS/BARON solver. An application has been performed in MKEK brass factory in KA +/- rA +/- kkale, Turkey and the solution of the application has been compared with alternative solution approaches based on cost and specification violation risk conditions. This comparison demonstrates that the proposed model is the most effective solution approach for managing stochastic uncertainties in blending problems and successfully can be used other industries such as alloy steel or secondary aluminum production.