AutoDock Vina is a new molecular docking and virtual screening program that significantly improves speed and accuracy compared to previous versions. It achieves a two-order-of-magnitude speed increase over AutoDock 4 while enhancing the accuracy of binding mode predictions. The program uses a new scoring function, efficient optimization, and multithreading to improve performance. It automatically calculates grid maps and clusters results without user intervention. The scoring function is based on a combination of steric, hydrophobic, and hydrogen bonding interactions. It uses a weighted sum of these interactions to predict binding affinity. The function is derived from the PDBbind database and is designed to be more accurate than previous empirical scoring functions. The optimization algorithm uses an iterated local search approach, which is efficient and effective for finding the global minimum of the scoring function. The program is designed to be user-friendly, with automatic grid map generation and result clustering. It is compatible with the PDBQT file format used in AutoDock 4, making it easy to integrate with existing tools. Vina does not have the same limitations as AutoDock 4, such as fixed maximum atom and rotatable bond numbers, allowing it to adapt to different input sizes. Vina's performance was tested on 190 protein-ligand complexes, showing a significant improvement in both speed and accuracy compared to AutoDock 4. It ran 62 times faster in single-threaded mode and 7.25 times faster with 8 threads. The RMSD values for predicted conformations were compared to experimental structures, with Vina showing better accuracy. The standard error of binding free energy predictions was also lower for Vina. The program is available online and is designed to be efficient and scalable, taking advantage of multi-core processors. Future work includes improving the scoring function and exploring the use of more cores for better performance. The software is supported by the NIH and has been developed using modern C++ techniques to ensure robustness, speed, and flexibility.AutoDock Vina is a new molecular docking and virtual screening program that significantly improves speed and accuracy compared to previous versions. It achieves a two-order-of-magnitude speed increase over AutoDock 4 while enhancing the accuracy of binding mode predictions. The program uses a new scoring function, efficient optimization, and multithreading to improve performance. It automatically calculates grid maps and clusters results without user intervention. The scoring function is based on a combination of steric, hydrophobic, and hydrogen bonding interactions. It uses a weighted sum of these interactions to predict binding affinity. The function is derived from the PDBbind database and is designed to be more accurate than previous empirical scoring functions. The optimization algorithm uses an iterated local search approach, which is efficient and effective for finding the global minimum of the scoring function. The program is designed to be user-friendly, with automatic grid map generation and result clustering. It is compatible with the PDBQT file format used in AutoDock 4, making it easy to integrate with existing tools. Vina does not have the same limitations as AutoDock 4, such as fixed maximum atom and rotatable bond numbers, allowing it to adapt to different input sizes. Vina's performance was tested on 190 protein-ligand complexes, showing a significant improvement in both speed and accuracy compared to AutoDock 4. It ran 62 times faster in single-threaded mode and 7.25 times faster with 8 threads. The RMSD values for predicted conformations were compared to experimental structures, with Vina showing better accuracy. The standard error of binding free energy predictions was also lower for Vina. The program is available online and is designed to be efficient and scalable, taking advantage of multi-core processors. Future work includes improving the scoring function and exploring the use of more cores for better performance. The software is supported by the NIH and has been developed using modern C++ techniques to ensure robustness, speed, and flexibility.