Measurements of long period disturbances associated with wave groups are compared with theoretical predictions. The experiments were carried out in a wave flume with an absorbing beach at one end. Tests were carried out using both regular wave groups (made up of two wave frequencies so that groups occur at a single beat frequency) and a random sea. The theoretical predictions are based on a Stokes expansion of the basic wave equations up to second order. The results are consistent with having three components in the long period disturbance at wave group periods. The first is set-down beneath wave groups, a disturbance that is tied to the wave groups travelling towards the beach. The second is a surf beat, a free long wave propagating back from the beach that is generated by reflection of the set-down from the beach. The third component is a secondary wave, a free long wave propagating towards the beach, that is generated by the wave-maker when it is only programmed to produce the primary wave system without its associated set-down beneath wave groups. The experimental results show that this third component can be minimised for both regular wave groups and a random sea by adding an appropriate long period movement of the wave-maker. Also, results are presented that indicate that long cycles of truly random waves, lasting many hours in model terms, are required to obtain reliable estimates of the effect of wave grouping on marine structures.