The paper introduces ParamILS, an automatic framework for optimizing the performance of algorithms by adjusting their parameters. The authors review local-search-based algorithm configuration procedures and present novel techniques to accelerate them by adaptively limiting the time spent on evaluating individual configurations. They demonstrate the effectiveness of these methods through comprehensive experimental evaluations, showing substantial and consistent performance improvements for both complete and incomplete algorithms, including the CPLEX mixed integer programming solver. The paper also discusses the application of ParamILS in various contexts, such as empirical studies, algorithm comparisons, and practical use, highlighting its ability to handle a wide range of parameter types and configurations. The authors conclude by summarizing the common patterns that contributed to ParamILS's success and provide practical advice for practitioners.The paper introduces ParamILS, an automatic framework for optimizing the performance of algorithms by adjusting their parameters. The authors review local-search-based algorithm configuration procedures and present novel techniques to accelerate them by adaptively limiting the time spent on evaluating individual configurations. They demonstrate the effectiveness of these methods through comprehensive experimental evaluations, showing substantial and consistent performance improvements for both complete and incomplete algorithms, including the CPLEX mixed integer programming solver. The paper also discusses the application of ParamILS in various contexts, such as empirical studies, algorithm comparisons, and practical use, highlighting its ability to handle a wide range of parameter types and configurations. The authors conclude by summarizing the common patterns that contributed to ParamILS's success and provide practical advice for practitioners.