This lecture discusses the estimation of frequency correlation on fading radio channels from time series of channel impulse response estimates derived from propagation measurements. The practical application for knowledge of frequency correlation is explained. Then, a text book derivation of the equation relating the channel's average power delay profile to its frequency correlation function is reviewed, with emphasis on explaining the minimum conditions needed for validity of the relationship. An alternative method, free from these conditions is also explained. Finally, examples are given, which apply the two methods to estimate frequency correlation on Rayleigh and Rician fading channels from data measured with a pseudo noise measurement system. Differences in results for the two channel types are highlighted and explained. An adjustment, reported by others, which is needed to correct results in order to compensate for the non-uniform shape of the measurement system transfer function is also demonstrated.