Most studies of the effects of predation on the organization and structure of prey communities have ignored predator population size structure and density. In the present investigation, we designed a series of field experiments to test the effects of these predator population properties on the intertidal benthic invertebrates of a salt marsh. The common mummichog, Fudulus heteroclitus, was the primary predator in our experiments. Three size—classes of mummichogs (small, 40—50) mm Total Length [TL]: medium, 50—69 mm TL; large >70 mm TL) were maintained at each of three densities (1, 2 and 4 fish/m2) in 10—m2 enclosures on the salt marsh. Water—filled pits were included in each cage as low—tide refuges for the fish. Each fish treatment combination was replicated three times. In addition, there were three uncaged 10—m2 plots (open controls) and three caged plots from which Fundulus were excluded (closed controls). Three cylindrical core samples (0.005 m2) were extracted from the sediment to a depth of 10 cm in each of the 33 experimental plots 1, 4, and 10 mo after initiation of the experiment in June 1978. Densities of the small, surface—dwelling invertebrates which make up most of the benthic community (Manayunkia aestuarina, Streblospio benedicti, Leptochelia rapax, ostracods, hydrobiid snails, and Gemma gemma) were highest in enclosures containing large fish, and lowest in closed controls and in enclosures containing small fish. Intermediate invertebrate densities occurred in enclosures containing medium—sized mumichogs. The effects of fish size were enhanced at high fish densities (2—4 fish/m2), although fish density alone had little effect on invertebrate abundance. We considered several hypotheses that could explain the findings. Small epibenthic predators—disturbers like the grass shrimp, Palemonetes pugio, were commonly observed in the experimental enclosures. The gut contents of fish taken from enclosures indicated that the shrimp were ingested only by medium and large size—classes of F. heteroclitus. We suggest that infaunal densities increased in response to the control of small epibenthic predator—disturbers (i.e., P. pugio) by the larger F. heteroclitus.