This Topic introduces the notion of estimating statistical averages from a single realisation of a stochastic process. This concept is most easily developed for estimating first and second moments of stationary random processes using time-averages. This requires the process to be Ergodic and wide-sense stationary (WSS). The video first introduces ergodicity from an intuitive perspective, and then further expands the definition in terms of using the properties of a consistent estimator. This is expressed through the two definitions of ergodic in the mean, or ergodic in correlation. Examples of non-ergodic and ergodic processes are presented. One very detailed example proves a process is ergodic in the mean through calculating the bias and variance of the time-average.