The feeding behaviour of eight species of medusae was studied; total food consumed, rates of digestion, and growth rate were noted, especially in Cyanea and Aurelia.Aurelia up to 5 cm diam can accept a wide range of size of food organisms obtained by filtering sea water through the curtain of tentacles during pulsation, whereas most other species find their food by random contact with the tentacles. In Bougainvillia it is the dichotomously branched oral tentacles that capture the food. The apparent selectivity of Sarsia in feeding on copepods is shown to be a physical effect produced by its method of swimming. Leuckartiara can catch and eat active organisms larger than itself. None of the medusae examined showed any reaction to artificial stimulation by a vibrating probe.The method of transfer of food from the tentacles to the mouth is closely associated with the structure of the medusa. Long trailing tentacles contract and lift food towards the mouth and if the manubrium is long it can be manipulated to pick off the food from the tentacles. If the tentacles are short the umbrella edge folds inwards towards the mouth. If there is no manubrium, as in Staurophora, the cruciform stomach can accept food at any point.Larval fish are important in the diet of most medusae and rough estimates are given of a probable consumption of 50–250 larval fish per hydromedusa, about 450–500 by each Aurelia, and in the order of 15,000 by Cyanea. A conversion factor is given for food eaten by medusae of 37%, though this is based on only few data.Following the work of Horridge and Boulton (Proc. Roy. Soc. (Ser. B) 168: 413–419, 1967), who found that Spadella could be artificially stimulated to bite vibrating probes, similar experiments were made with Sagitta elegans. Although the chaetognaths were successfully kept alive the probes produced no feeding reaction. The methods are outlined for the benefit of others wishing to continue such investigations.