The paper opens with a review of the theoretical studies of coastal deviation which have been made in recent years. It shows that there is close agreement between the various formulae derived for the deviation and that the graphical method used in an earlier paper by two of the present authors produces results which are substantially in agreement with those obtained from the formulae.In an endeavour to obtain experimental confirmation of the theoretical magnitude of the deviation and its variation with distance from the boundary with angle of incidence of the wave front at the boundary, etc., existing data from various sources have been examined. Two new series of measurement on low- and medium-frequency waves have also been made. In both cases the transmitters were located at sea and the directional measurements taken at numerous sites on land. The general conclusion obtained from this experimental investigation was that a deviation due solely to the change in ground conductivity at the boundary and of the magnitude given by theory is small compared with the larger random directional errors which are attributed to ground irregularities, surface obstructions and other causes. The only systematic deviations observed had amplitudes of up to 7° and occurred within a few hundred metres (less than a half-wavelength) of the coast on the seaward side; at such near distances the theoretical formulae are not valid. It is concluded that for many coastal direction-finding stations coastal deviation as such is unimportant.