Shoreline variability at embayed beaches can be characterized into modes where either longshore or cross-shore sediment transport processes dominate the overall shoreline response, or there is a mixed combination of the two. To-date it has been assumed that the relative dominance of these differing modes of longshore and/or cross-shore shoreline behaviour is stationary in time. This concept is tested using a unique 43-year dataset of shoreline positions at Narrabeen-Collaroy beach (southeast Australia) and a rolling five-year window Empirical Orthogonal Function analysis, revealing the new observation of a distinct interannual variability in the dominant cross-shore and longshore modes of shoreline behaviour at this site. The dominant mode of shoreline behaviour was found to range from time periods when the cross-shore mode (referred to as the cross-shore coherent mode) comprised as much as 74 % of the overall shoreline variability, contrasting to other time periods when the alongshore mode of shoreline behaviour (longshore coherent mode) was more dominant, accounting for up to 62 % of the observed shoreline variability. Wave forcing correlation analysis suggests that these modes are controlled by varying influences of wave intensity and wave direction at interannual time scales. Consistent with previous research at this same embayment, the cross-shore coherent mode of shoreline variability appears to be controlled primarily by wave height/intensity, with stronger controls (i.e., higher correlation) when this cross-shore mode was overwhelmingly dominant. In contrast, the contribution of the longshore coherent mode appears to be controlled primarily by wave direction, but also at certain unique times in the time series by wave intensity. Analysis using available topographic and bathymetric data suggests that the observed switch in longshore versus cross-shore dominance may be triggered by extreme storm events, which cause significant and near ‘instantaneous’ redistribution of sediment across the entire shoreface and beach face. These results highlight the importance of considering a non-stationary shoreline behaviour at embayed beaches and the association of differing modes of dominant shoreline behaviour with interannual wave climate variability. Given observed interannual variability in deep water wave climates more broadly, it is likely that the dominant modes of shoreline variability may also occur at other embayed beaches and should be considered for numerical modelling (and prediction) of future shoreline behaviour.