The paper introduces the MCMCglmm R package, which implements Markov chain Monte Carlo (MCMC) methods for fitting multi-response generalized linear mixed models. These models are flexible and can handle non-Gaussian response variables, allowing for the analysis of multiple response variables simultaneously. The package supports a wide range of distributions, including Gaussian, Poisson, binomial, exponential, zero-inflated, and censored distributions. It also allows for various variance structures for random effects and residuals, including interactions with categorical or continuous variables and more complex structures arising from shared ancestry. The software is designed to be user-friendly and efficient, with all simulation performed in C/C++ using the CSparse library for sparse linear systems. The paper includes a detailed explanation of the model structure, parameter estimation, and the use of the MCMCglmm function, along with a worked example from quantitative genetics. The package is compared with WinBUGS, showing significantly faster computation and higher effective sample sizes. The paper concludes by highlighting the package's potential to make advanced mixed modeling techniques more accessible to researchers.The paper introduces the MCMCglmm R package, which implements Markov chain Monte Carlo (MCMC) methods for fitting multi-response generalized linear mixed models. These models are flexible and can handle non-Gaussian response variables, allowing for the analysis of multiple response variables simultaneously. The package supports a wide range of distributions, including Gaussian, Poisson, binomial, exponential, zero-inflated, and censored distributions. It also allows for various variance structures for random effects and residuals, including interactions with categorical or continuous variables and more complex structures arising from shared ancestry. The software is designed to be user-friendly and efficient, with all simulation performed in C/C++ using the CSparse library for sparse linear systems. The paper includes a detailed explanation of the model structure, parameter estimation, and the use of the MCMCglmm function, along with a worked example from quantitative genetics. The package is compared with WinBUGS, showing significantly faster computation and higher effective sample sizes. The paper concludes by highlighting the package's potential to make advanced mixed modeling techniques more accessible to researchers.