The mirt package is an R package designed for estimating parameters in multidimensional item response theory (IRT) models, including both exploratory and confirmatory models. It uses maximum-likelihood methods and provides tools for estimating item parameters, factor scores, and model fit statistics. The package addresses limitations of existing R packages like ltm and eRm, which are restricted to unidimensional IRT models, and MCMCpack, which requires Bayesian diagnostics and is computationally intensive. The mirt package supports a wide range of IRT models, including the three-parameter logistic model, ordinal response model, and bifactor models. It also includes stochastic estimation techniques such as the Metropolis-Hastings Robbins-Monro algorithm for parameter estimation. The package is implemented with functions for exploratory and confirmatory item analysis, and it provides diagnostic tools for model evaluation. The mirt package is efficient for high-dimensional models and offers flexibility in specifying model constraints. It is compared with other IRT software in terms of estimation accuracy and computational efficiency, showing superior performance in many cases. The package is actively developed and includes future features such as limited-information model fit statistics and multiple-group estimation.The mirt package is an R package designed for estimating parameters in multidimensional item response theory (IRT) models, including both exploratory and confirmatory models. It uses maximum-likelihood methods and provides tools for estimating item parameters, factor scores, and model fit statistics. The package addresses limitations of existing R packages like ltm and eRm, which are restricted to unidimensional IRT models, and MCMCpack, which requires Bayesian diagnostics and is computationally intensive. The mirt package supports a wide range of IRT models, including the three-parameter logistic model, ordinal response model, and bifactor models. It also includes stochastic estimation techniques such as the Metropolis-Hastings Robbins-Monro algorithm for parameter estimation. The package is implemented with functions for exploratory and confirmatory item analysis, and it provides diagnostic tools for model evaluation. The mirt package is efficient for high-dimensional models and offers flexibility in specifying model constraints. It is compared with other IRT software in terms of estimation accuracy and computational efficiency, showing superior performance in many cases. The package is actively developed and includes future features such as limited-information model fit statistics and multiple-group estimation.