The paper discusses the development and application of AutoDock 2.4, a computational tool for predicting the binding conformations of flexible ligands to nonflexible macromolecular targets. The technique combines simulated annealing (SA) for conformation search with a grid-based energy evaluation method based on the AMBER force field. AutoDock 2.4 has been optimized for performance and accuracy, incorporating enhancements such as an intuitive interface. The authors have also developed tools for parallel execution of multiple docking jobs on a heterogeneous network of UNIX-based workstations. The paper presents results from a suite of diverse test systems, demonstrating that even for ligands with many degrees of freedom, root-mean-square deviations of less than 1 Å from the crystallographic conformation can be achieved for the lowest-energy dockings. The introduction highlights the differences between 'unbiased' and 'directed' docking methods, with AutoDock being an example of the former. The materials and methods section details the docking protocols, including the use of random initialization and maximum allowable initial energy to avoid steric clashes and ensure computational feasibility.The paper discusses the development and application of AutoDock 2.4, a computational tool for predicting the binding conformations of flexible ligands to nonflexible macromolecular targets. The technique combines simulated annealing (SA) for conformation search with a grid-based energy evaluation method based on the AMBER force field. AutoDock 2.4 has been optimized for performance and accuracy, incorporating enhancements such as an intuitive interface. The authors have also developed tools for parallel execution of multiple docking jobs on a heterogeneous network of UNIX-based workstations. The paper presents results from a suite of diverse test systems, demonstrating that even for ligands with many degrees of freedom, root-mean-square deviations of less than 1 Å from the crystallographic conformation can be achieved for the lowest-energy dockings. The introduction highlights the differences between 'unbiased' and 'directed' docking methods, with AutoDock being an example of the former. The materials and methods section details the docking protocols, including the use of random initialization and maximum allowable initial energy to avoid steric clashes and ensure computational feasibility.