The brms package in R implements Bayesian multilevel models using the probabilistic programming language Stan. It supports a wide range of distributions and link functions, enabling users to fit linear, robust linear, binomial, Poisson, survival, ordinal, zero-inflated, hurdle, and non-linear models in a multilevel context. It also allows for autocorrelation, user-defined covariance structures, censored data, and meta-analytic standard errors. Prior specifications are flexible and encourage users to apply priors reflecting their beliefs. Model fit can be assessed using the Watanabe-Akaike information criterion and leave-one-out cross-validation. The paper introduces brms, which allows users to fit Bayesian multilevel models using simple, lme4-like formula syntax. It supports a wide range of distributions and link functions, multiple grouping factors, autocorrelation, user-defined covariance structures, and flexible prior specifications. The package is available from CRAN and can be installed via R or GitHub. It uses Stan for model fitting and provides functions for model specification, data preparation, and posterior analysis. The paper discusses the structure of multilevel models, the use of Stan for Bayesian inference, and the advantages of brms over other R packages. It compares brms with other packages like lme4, MCMCglmm, rstanarm, and rethinking, highlighting its flexibility in handling various model types, including ordinal and zero-inflated models. It also discusses prior specifications, model fitting, and posterior analysis, including the use of WAIC and LOO for model comparison. The paper provides examples of using brms to fit models to real data, such as infection recurrence times in kidney patients and inhaler instruction ratings. It explains the syntax for model specification, the use of priors, and the interpretation of results. The paper concludes that brms is a powerful tool for Bayesian multilevel modeling, offering flexibility, ease of use, and robustness in handling complex models. Future plans include expanding the package to support multivariate models, distributional regression models, and mixture response distributions.The brms package in R implements Bayesian multilevel models using the probabilistic programming language Stan. It supports a wide range of distributions and link functions, enabling users to fit linear, robust linear, binomial, Poisson, survival, ordinal, zero-inflated, hurdle, and non-linear models in a multilevel context. It also allows for autocorrelation, user-defined covariance structures, censored data, and meta-analytic standard errors. Prior specifications are flexible and encourage users to apply priors reflecting their beliefs. Model fit can be assessed using the Watanabe-Akaike information criterion and leave-one-out cross-validation. The paper introduces brms, which allows users to fit Bayesian multilevel models using simple, lme4-like formula syntax. It supports a wide range of distributions and link functions, multiple grouping factors, autocorrelation, user-defined covariance structures, and flexible prior specifications. The package is available from CRAN and can be installed via R or GitHub. It uses Stan for model fitting and provides functions for model specification, data preparation, and posterior analysis. The paper discusses the structure of multilevel models, the use of Stan for Bayesian inference, and the advantages of brms over other R packages. It compares brms with other packages like lme4, MCMCglmm, rstanarm, and rethinking, highlighting its flexibility in handling various model types, including ordinal and zero-inflated models. It also discusses prior specifications, model fitting, and posterior analysis, including the use of WAIC and LOO for model comparison. The paper provides examples of using brms to fit models to real data, such as infection recurrence times in kidney patients and inhaler instruction ratings. It explains the syntax for model specification, the use of priors, and the interpretation of results. The paper concludes that brms is a powerful tool for Bayesian multilevel modeling, offering flexibility, ease of use, and robustness in handling complex models. Future plans include expanding the package to support multivariate models, distributional regression models, and mixture response distributions.