The Bond rod mill grindability test is a widely used tool to estimate the response of ores to rod milling. However, the mass of ore required to conduct the test, which can typically vary from about 8 to as much as 20 kg, may represent a challenge when sample availability is critical. The present work proposes an alternative procedure, which is based on the size-mass balance model with a specific account for the acceleration of breakage rates of the coarsest size fraction, a phenomenon that is well documented to occur during rod milling. The model was validated for four materials, showing that it can predict the Bond rod mill work index with deviations below 10%. Requiring only 1.5-3 kg of sample, the procedure is demonstrated to be a viable alternative to the standard test, whenever only a limited amount of sample is available. (C) 2011 Elsevier Ltd. All rights reserved.