A three-dimensional computational fluid dynamics (CFD) model is developed to investigate the influence of geometric and operating parameters on performance of a direct methanol fuel cell (DMFC). Semi-empirical relationships are introduced to describe the electrochemical behaviors required in the CFD governing equations. Coefficients in these semi-empirical relationships are fitted using experimental data. Two geometric configurations with serpentine channels at the anode and cathode are considered in this work. Temperature, methanol concentration, and methanol flow rate are selected as the operating parameters. Due to the computational effort of CFD, an adaptive metamodeling method is developed to reduce the number of data-fitting iterations for obtaining the coefficients in the semi-empirical relationships. The effectiveness of the method is demonstrated by fitting the model using the experimental data collected from the first geometric configuration of the DMFC and comparing the predicted performance of the second configuration with its experimental performance. A commercial CFD system, Fluent 12.0, was used in this research. Copyright (C) 2013, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.