Automated docking of substrates to proteins of known structure aids the process of crystallographic analysis in two ways. First, automated docking can be used to generate a small number of starting models for substrates using only protein coordinates from an early stage of refinement. Second, automated docking provides a method for exploring aspects of catalysis that are inaccessible to crystallography by postulating binding modes of catalytic intermediates. This paper describes the use of automated docking to explore the binding of substrates to aconitase. The technique starts with a substrate molecule in an arbitrary configuration and position and finds favorable docked configurations in a (static) protein active site based on a molecular mechanics type force field. Using protein coordinates from an early stage of refinement of an aconitase-isocitrate complex, we successfully predicted the binding configuration of isocitrate. Four configurations were found, the energetically most favorable of which fit the observed electron density well and was used as a starting model for further refinement. Two configurations were found in citrate docking experiments, the second of which approximates the mode of substrate binding in an aconitase-nitrocitrate complex. We were also able to propose two binding modes of the catalytic intermediate cis-aconitate. These correspond closely to the isocitrate and the citrate binding modes. The relation of these new results to the proposed reaction mechanism is discussed.