Functional magnetic resonance imaging (fMRI) data is usually analyzed using hemodynamic response data formulated with the aid of a gamma function. An fMRI response to stimuli can be mathematically modeled by convolution of the hemodynamic response and the presented stimulus. This approach is based on a linear system analysis. However, it is known that most biological systems are nonlinear and the nature of the hemodynamic response depends on both the subjects and brain regions under study. In this work, we estimated the hemodynamic responses of the fusiform face area (FFA) and the primary visual area (VI) using 1st- and 2nd-order Volterra kernels. The estimated hemodynamic responses were used to analyze fMRI data obtained from the corresponding regions. The results of the analysis of fMRI data using the estimated hemodynamic responses show a more significant activation than is obtained by conventional analysis using the hemodynamic response modeled by gamma function. (C) 2004 Published by Elsevier B.V.