The threshold of movement of sediment obtained from sandbanks within the Bristol Channel (UK) is investigated under unidirectional, oscillatory and combined flows. The experiments were undertaken in a recirculating, unidirectional laboratory flume containing an oscillating plate to simulate wave action, with movement along the same axis as the unidirectional flows. The sand samples consisted of cohesionless quartz grains with median grain sizes between 0·315 and 0·513 mm. The experiments were performed under flow velocities (measured at 2 cm above the bed) ranging between 0 and 24 cm s–1 and oscillatory currents (wave periods of 5, 12 and 15 s) ranging from 0 to 28 cm s–1. The critical conditions for the initiation of sediment movement were assessed, by visual observation, using the Yalin criterion. The results show that, under unidirectional flow, there is a slight overestimation of the threshold of naturally graded sediments derived on the basis of empirically derived threshold curves for artificially prepared sediments under similar flow conditions. In the case of oscillatory flows, the threshold for the natural sands is found to be higher than that predicted by previously derived empirical curves. Under combined flows, wave period is shown to control threshold conditions, with the unidirectional and oscillatory flow components combining in a linear fashion for long-period (12 s and 15 s) waves. In contrast, in the presence of short-period (5 s) waves, the unidirectional and oscillatory components of the flow appear to ‘decouple’. For high orbital velocities, in both cases, the effect of the wave period on threshold diminishes.