This paper introduces the COMBINE software package, a statistical analysis tool used by the CMS Collaboration for high-energy physics experiments. Originally designed for Higgs boson searches, COMBINE has evolved to support a wide range of statistical analyses and measurements. The package is built around ROOT, ROOFIT, and ROOSTATS, and is designed to provide a command-line interface for common workflows and encapsulate statistical models in human-readable configuration files, known as "datacards." COMBINE allows for the combination of likelihoods from independent datasets to enhance sensitivity in searches or measurements. The paper outlines the installation process, dependencies, and the statistical model construction, which factors the primary and auxiliary observables into manageable components. It details the supported analysis types, including counting and shape analyses, and provides examples of datacards for each type. The counting analysis focuses on event counts, while the shape analysis incorporates more complex observables, such as histograms or parametric functions. The paper also explains how to handle systematic uncertainties and the use of template-based and parametric shape analyses. COMBINE's capabilities are demonstrated through examples, and the paper emphasizes its versatility and adaptability for various statistical analyses in high-energy physics. The online documentation is recommended for comprehensive and up-to-date information.This paper introduces the COMBINE software package, a statistical analysis tool used by the CMS Collaboration for high-energy physics experiments. Originally designed for Higgs boson searches, COMBINE has evolved to support a wide range of statistical analyses and measurements. The package is built around ROOT, ROOFIT, and ROOSTATS, and is designed to provide a command-line interface for common workflows and encapsulate statistical models in human-readable configuration files, known as "datacards." COMBINE allows for the combination of likelihoods from independent datasets to enhance sensitivity in searches or measurements. The paper outlines the installation process, dependencies, and the statistical model construction, which factors the primary and auxiliary observables into manageable components. It details the supported analysis types, including counting and shape analyses, and provides examples of datacards for each type. The counting analysis focuses on event counts, while the shape analysis incorporates more complex observables, such as histograms or parametric functions. The paper also explains how to handle systematic uncertainties and the use of template-based and parametric shape analyses. COMBINE's capabilities are demonstrated through examples, and the paper emphasizes its versatility and adaptability for various statistical analyses in high-energy physics. The online documentation is recommended for comprehensive and up-to-date information.