The paper introduces the R package `geepack`, which implements the Generalized Estimating Equations (GEE) approach for fitting marginal generalized linear models to clustered data. GEE is particularly useful for analyzing longitudinal data and repeated measures where observations within a cluster are correlated but observations in separate clusters are independent. The package offers several key features, including an interface similar to `glm`, a jackknife variance estimator, and the ability to incorporate covariates into the scale and correlation parameters. The paper illustrates these features through an example dataset on respiratory illness, demonstrating how to fit a logistic model using GEE and compare different working correlation structures. The analysis highlights the importance of accounting for within-patient correlation and the robustness of the GEE approach in handling small numbers of clusters. The conclusion emphasizes the flexibility and practicality of `geepack` for estimating covariate-dependent mean, scale, and correlation parameters in correlated observations.The paper introduces the R package `geepack`, which implements the Generalized Estimating Equations (GEE) approach for fitting marginal generalized linear models to clustered data. GEE is particularly useful for analyzing longitudinal data and repeated measures where observations within a cluster are correlated but observations in separate clusters are independent. The package offers several key features, including an interface similar to `glm`, a jackknife variance estimator, and the ability to incorporate covariates into the scale and correlation parameters. The paper illustrates these features through an example dataset on respiratory illness, demonstrating how to fit a logistic model using GEE and compare different working correlation structures. The analysis highlights the importance of accounting for within-patient correlation and the robustness of the GEE approach in handling small numbers of clusters. The conclusion emphasizes the flexibility and practicality of `geepack` for estimating covariate-dependent mean, scale, and correlation parameters in correlated observations.