beta-Sialon powders were synthesized by microwave-assisted carbothermal reduction of powder materials resultant from a sol-gel process using Si(OC2H5)(4), Al(NO3)(3)center dot 9H(2)O and sucrose (C12H22O11) as the main starting materials and artificial neural networks (ANNs) was applied to model and predict relative contents of beta-Sialon in the final product samples. beta-Sialon was formed at as low as 1250 degrees C by using the technique developed with the present work. Furthermore, addition of Fe2O3 promoted the beta-Sialon formation. As-prepared beta-Sialon ultrafine powders were granular with primary size of about 69 nm. A back propagation (BP) ANNs was used to establish a model to predict the reaction extents (relative contents of beta-Sialon) under various processing conditions. The results indicated that the BP ANNs could be effectively used to establish the nonlinear relationships between the relative contents of beta-Sialon and the processing conditions. (C) 2015 The Society of Powder Technology Japan. Published by Elsevier B.V. and The Society of Powder Technology Japan. All rights reserved.