The *mirt* package is a comprehensive tool for estimating multidimensional item response theory (IRT) parameters, addressing the limitations of existing R packages such as ltm, eRm, and MCMCpack. It supports both exploratory and confirmatory IRT models using maximum-likelihood methods, including the Gauss-Hermite quadrature method for EM estimation and the Metropolis-Hastings Robbins-Monro (MH-RM) algorithm for stochastic estimation. The package offers functions like `mirt()`, `bfactor()`, `polymirt()`, and `confmirt()` for estimating various IRT models, with detailed diagnostics and plotting features. Compared to other R packages, *mirt* is more efficient and accurate, especially for higher-dimensional models. Future developments include adding limited-information model fit statistics, standard errors for EM solutions, multiple-group estimation, and nominal and rating scale intercept methods for polytomous data.The *mirt* package is a comprehensive tool for estimating multidimensional item response theory (IRT) parameters, addressing the limitations of existing R packages such as ltm, eRm, and MCMCpack. It supports both exploratory and confirmatory IRT models using maximum-likelihood methods, including the Gauss-Hermite quadrature method for EM estimation and the Metropolis-Hastings Robbins-Monro (MH-RM) algorithm for stochastic estimation. The package offers functions like `mirt()`, `bfactor()`, `polymirt()`, and `confmirt()` for estimating various IRT models, with detailed diagnostics and plotting features. Compared to other R packages, *mirt* is more efficient and accurate, especially for higher-dimensional models. Future developments include adding limited-information model fit statistics, standard errors for EM solutions, multiple-group estimation, and nominal and rating scale intercept methods for polytomous data.