In this chapter we discuss the generalized estimating equations (GEE) approach for analyzing longitudinal data. Over the past 20 years, the GEE approach has proven to be an exceedingly useful method for the analysis of longitudinal data, especially when the response variable is discrete (e.g., binary, ordinal, or a count). When the longitudinal response is discrete, linear models (e.g., linear mixed-effects models) are not very appealing for relating changes in the mean response to covariates for at least two main reasons. First, with a discrete response there is intrinsic dependence of the variability on the mean. Second, the range of the mean response (e.g., a proportion or rate for a response that is binary or a count, respectively) is constrained. In the setting of regression modeling of a univariate response, both of these aspects of the response can be conveniently accommodated within generalized linear models via known variance and link functions.