This study investigates the predictive ability of gene-expression programming (GEP) in the estimation of methane yield (Y-m) and effluent substrate (S-e) produced by two anaerobic filters. The modeling study was carried out using the data obtained from two upflow anaerobic filters - one mesophilic (35 degrees C) and one thermophilic (55 degrees C) - operated for the treatment of paper-mill wastewater under varying organic loadings. The GEP model was composed of three inputs, hydraulic retention time (T-hr), organic loading rate (R-ol), and influent substrate (S-i), and one output, either S-e or Y-m. The Stover-Kincannon model was also used for data analysis and to evaluate the prediction ability. Three statistical criteria, root mean square error (RMSE), determination coefficient (R-2), and Akaike's information criteria (AIC), were the means used for comparison. The results showed that the GEP approach predicted the performance of both anaerobic filters much better than the Stover-Kincannon model. Crown Copyright (C) 2010 Published by Elsevier B.V. All rights reserved.