The irace package implements iterated racing for automatic algorithm configuration. It is a software tool that automates the process of finding optimal parameter settings for optimization algorithms. The package includes iterated racing procedures, which have been successfully used to configure various state-of-the-art algorithms. The iterated racing procedures in irace include the iterated F-race algorithm and several extensions and improvements over it. The paper describes the rationale behind iterated racing procedures and introduces recent extensions, including a restart mechanism to avoid premature convergence, the use of truncated sampling distributions to handle parameter bounds, and an elitist racing procedure to ensure that the best configurations are evaluated on the highest number of training instances. The paper also evaluates the most recent version of irace and demonstrates its use and potential in various example applications. The irace package is implemented as a general iterated racing procedure that includes I/F-Race as a special case. It also implements several extensions, such as the use of the paired t-test instead of Friedman's test. The package incorporates several improvements, such as sampling from a truncated normal distribution, a parallel implementation, a restart strategy, and an elitist racing procedure. The paper is structured as follows: Section 2 introduces the algorithm configuration problem and gives an overview of approaches to automatic algorithm configuration. Section 3 describes the iterated racing procedure as implemented in the irace package and several further extensions. Section 4 illustrates the steps followed to apply irace to two configuration scenarios and compares experimentally the elitist and non-elitist variants. In Section 5, the paper gives an overview of articles that have used irace for configuration tasks. The paper concludes in Section 6. Appendix A provides a brief description of the irace package itself, its components, and its main options. The paper discusses the algorithm configuration problem, which involves finding good parameter settings for solving unseen problem instances by learning on a set of training problem instances. The paper describes the iterated racing algorithm as implemented in the irace package, which includes a restart mechanism to avoid premature convergence, the use of truncated sampling distributions, and an elitist racing procedure. The paper also discusses other features of irace, such as initial configurations and parallel evaluation of configurations. The paper concludes that the irace package is a powerful tool for automatic algorithm configuration and has the potential to lead to new design paradigms for optimization software.The irace package implements iterated racing for automatic algorithm configuration. It is a software tool that automates the process of finding optimal parameter settings for optimization algorithms. The package includes iterated racing procedures, which have been successfully used to configure various state-of-the-art algorithms. The iterated racing procedures in irace include the iterated F-race algorithm and several extensions and improvements over it. The paper describes the rationale behind iterated racing procedures and introduces recent extensions, including a restart mechanism to avoid premature convergence, the use of truncated sampling distributions to handle parameter bounds, and an elitist racing procedure to ensure that the best configurations are evaluated on the highest number of training instances. The paper also evaluates the most recent version of irace and demonstrates its use and potential in various example applications. The irace package is implemented as a general iterated racing procedure that includes I/F-Race as a special case. It also implements several extensions, such as the use of the paired t-test instead of Friedman's test. The package incorporates several improvements, such as sampling from a truncated normal distribution, a parallel implementation, a restart strategy, and an elitist racing procedure. The paper is structured as follows: Section 2 introduces the algorithm configuration problem and gives an overview of approaches to automatic algorithm configuration. Section 3 describes the iterated racing procedure as implemented in the irace package and several further extensions. Section 4 illustrates the steps followed to apply irace to two configuration scenarios and compares experimentally the elitist and non-elitist variants. In Section 5, the paper gives an overview of articles that have used irace for configuration tasks. The paper concludes in Section 6. Appendix A provides a brief description of the irace package itself, its components, and its main options. The paper discusses the algorithm configuration problem, which involves finding good parameter settings for solving unseen problem instances by learning on a set of training problem instances. The paper describes the iterated racing algorithm as implemented in the irace package, which includes a restart mechanism to avoid premature convergence, the use of truncated sampling distributions, and an elitist racing procedure. The paper also discusses other features of irace, such as initial configurations and parallel evaluation of configurations. The paper concludes that the irace package is a powerful tool for automatic algorithm configuration and has the potential to lead to new design paradigms for optimization software.