The MixSIAR model is an advanced mixing model that integrates various improvements from previous software packages (e.g., IsoSource, MixSIR, SIAR, IsotopeR) into a unified framework. It uses the write_JAGS_model function to generate a JAGS model file based on user input, allowing for flexible modeling of source data and mixture data. The model supports three options for fitting source data: raw data with covariance, summary statistics (mean, SD, n) with no covariance, and mean/SD/n with large sample sizes to fix source means. For raw source data, the model fits hierarchical data with normal distributions and covariance matrices, using diffuse priors for source means and variances. For summary statistics, the model estimates source means and variances using normal and chi-squared distributions. The model also allows for fixed, random, and continuous effects, with fixed effects being offsets from the overall intercept in ILR-space. Random effects are drawn from a shared distribution with mean zero. Continuous effects are added as linear terms in ILR-space. The model calculates mixture means as convex combinations of source proportions and fitted source means, adjusted by TDF and tracer concentration. Mixture variance can be modeled in three ways: process x residual error (with or without covariance), residual error only, or process error only. The model uses Wishart or inverse Wishart priors for variance components, depending on the error structure. The likelihood is based on normal distributions with mixture means and covariances. MixSIAR is designed to handle compositional data, such as fatty acid profiles, using the Central Limit Theorem and alternative software when necessary. The model is modular and recommended for inclusion in publications as it specifies the equations used.The MixSIAR model is an advanced mixing model that integrates various improvements from previous software packages (e.g., IsoSource, MixSIR, SIAR, IsotopeR) into a unified framework. It uses the write_JAGS_model function to generate a JAGS model file based on user input, allowing for flexible modeling of source data and mixture data. The model supports three options for fitting source data: raw data with covariance, summary statistics (mean, SD, n) with no covariance, and mean/SD/n with large sample sizes to fix source means. For raw source data, the model fits hierarchical data with normal distributions and covariance matrices, using diffuse priors for source means and variances. For summary statistics, the model estimates source means and variances using normal and chi-squared distributions. The model also allows for fixed, random, and continuous effects, with fixed effects being offsets from the overall intercept in ILR-space. Random effects are drawn from a shared distribution with mean zero. Continuous effects are added as linear terms in ILR-space. The model calculates mixture means as convex combinations of source proportions and fitted source means, adjusted by TDF and tracer concentration. Mixture variance can be modeled in three ways: process x residual error (with or without covariance), residual error only, or process error only. The model uses Wishart or inverse Wishart priors for variance components, depending on the error structure. The likelihood is based on normal distributions with mixture means and covariances. MixSIAR is designed to handle compositional data, such as fatty acid profiles, using the Central Limit Theorem and alternative software when necessary. The model is modular and recommended for inclusion in publications as it specifies the equations used.