This chapter presents a variety of methods for meta-analyses in epidemiology where first, the suitability of meta-analyses in epidemiological research is discussed. Next, the necessary steps to conduct a meta-analysis are shown. Then, statistical methods for the conduct of a meta-analysis are presented, where we start with the description of the estimation of a summary effect together with a classical test for heterogeneity. Here, Cochran’s Q-test and the I2 measure for heterogeneity between studies are introduced. We then show that meta-analyses can be performed in the context of linear mixed effects models of which the common effect (fixed) model is a special case. An extension of this model including random effects allows modeling of heterogeneity between studies. This heterogeneity is quantified by estimating the heterogeneity variance. Several methods to estimate this heterogeneity variance are introduced including the DerSimonian-Laird, maximum and restricted maximum likelihood as well as finite mixture models. If heterogeneity is present, a meta-regression can be used for explanation using study-specific covariates. Performing a meta-analysis one has to be aware of a major bias that may occur when statistically significant results are more likely to be published than non-significant ones. Several methods for the assessment of publication bias are introduced as well as a method which allows to adjust for publication bias. These methods are illustrated using an example based on a meta-analysis dealing with aspirin as a promising agent for the chemoprevention of lung cancer. Finally, based on these data the code for the freely available software R is provided to reproduce the calculations in this article.