Determining the strengths of interactions among species in natural communities presents a major challenge to ecology. Using an approach combining experimental perturbations and path analysis, I examined the mechanisms by which birds directly and indirectly affected other members of an intertidal community, evaluated alternative causal hypotheses, and predicted whether interactions among the unmanipulated species would be strong or weak. Comparing treatments with t tests indicated that excluding bird predators with cages caused increases in Pollicipes polymerus, and declines in Nucella spp., Mytilus californianus, and Semibalanus cariosus. However, these conclusions provided no insight into the underlying mechanisms causing the differences. Path analysis permitted insight into the causal mechanisms by making a variety of predictions about the strength of direct interactions: (a) Bird predation negatively affects Pollicipes, but not Nucella, Leptasterias, or Mytilus; (b) Pollicipes reduces Semibalanus and Mytilus abundance because of space competition; (c) Mytilus reduces Semibalanus cover through competition for space; and (d) as prey species, Semibalanus and Pollicipes enhance Nucella density, but Nucella predation does not have important effects on Semibalanus of Pollicipes. Based on the estimated strength of direct interactions, the importance of indirect effects among species could also be predicted. In experiments manipulating Nucella, Pollicipes, Semibalanus, and birds independently of one another, I tested 11 of the interactions predicted by the path analysis; all were supported. Path analysis in conjunction with limited experiments may provide an efficient means to predict important direct and indirect interactions among unmanipulated species within ecological communities.