Variation in the seasonal intensity of weather systems generating onshore winds incident on Port Kembla, N.S.W., have been examined for the period 1965 to 1983. The onshore-wind events were identified from anemograph records. In turn, they were related to synoptic weather records. An intensity index (1) for each weather system was estimated as a function of the barometric pressure gradient between Port Kembla and the system centre. The events were grouped by seasons and the intensity of the upper-tercile ( I 1 3 ) value for each season, utilised to represent the record for that season. Values of the seasonal I 1 3 intensities were spectrally analysed by Fast Fourier Transform techniques. The results identified quasi-periodic variations of onshore wind activity with recurrence intervals corresponding to those in the frequency band shared by the Southern Oscillation (SO), as well as the stratospheric Quasi-Biennial Oscillation (QBO) and the tropospheric quasi-biennial oscillation (BO). The dominant quasi-periodic components, identified from a time series filtered to remove high frequency fluctuations, include 3 years, 27 months and 21.6 months. The triennial and biennial components combine to account for approximately 32% of variance in the time series describing seasonal I 1 3 anomalies in onshore wind activity. The filtered time series correlated with the Southern Oscillation Index (SOI) ∗ ∗ Angell (1981). , with r = 0.49, p < 0.0001. The co-spectra for 14 years of the two series were examined for coherence and found to be 99% significant for each, at 2.2 and 3.5 years. The phase relationship between the two time series indicated that the SOI led the onshore-wind intensity by approximately one season. (This is the limit of accuracy afforded by the wind data.) A good correspondence of the I 1 3 anomalies with the period of both the QBO and the M1 index of the tropospheric biennial oscillation were also noted. In the latter instance, direct comparison of the M1 phase shift ∗∗ ∗∗ Trenberth (1980). was not possible because the I 1 3 data is contaminated with a strong SOI type signal. Peak I 1 3 - intensities are in phase to ±1 month with the monthly mean-sea-level peaks at Fort Denison, Sydney, N.S.W., over corresponding periods. More accurate resolution is not possible because of the specification of the weather system intensities as seasonal values. In turn, the sea-level fluctuations lead movement of sediment on Warilla Beach, N.S.W., by a month. Over the 9 years for which observations are available from 1976 to 1985, peak weather-system intensities for Port Kembla are co-incidental with low volumes of sediment stored above MSL on Warilla Beach. Conversely, periods of low onshore-wind activity are associated with increases in the volume of sediment stored on the beach.