In many research areas such as medical, psychological or social sciences, various questions arise on the causal relationships between different observed features. One aim of causal inference is to estimate the causal effect of some variable(s) on a certain outcome variable. This case applies, for instance, when we study the causal effect of a treatment for an individual via potential outcomes. The fundamental problem in causal inference is that, we only observe a single outcome for every individual, instead of all possible outcomes. Therefore we cannot compute the causal effect (estimator) from the data alone, i.e. without using prior knowledge on the data-generating process and without assessing the conditional treatment distribution conditional on the covariates. In this talk we will present the notion of average treatment effect (ATE), review several estimators for ATE and discuss their consistency, optimality and robustness in different settings (in randomized trials, in the presence of confounders).