Published results from caging experiments in soft sediments give the impression that predators generally have large effects on prey communities. However, the extent to which these results accurately reflect the prevalence of strong trophic links in benthic food webs is open to debate. This article examines the value of predator-caging experiments for testing the strength of trophic links in communities and considers the role that energy-flow studies or energy-flow models can play. We conclude that energy-flow studies or energetic modeling is insufficient to establish unequivocally the functional role of predators. Although such studies may be suggestive of strong controlling links, when experiments are performed, the results often contradict the conclusions from energetic analyses. The statistical power of caging experiments is assessed, and the potential for prey movement to confound experimental results is discussed. It is concluded that manipulative experiments still remain the only valid test of the predation hypothesis but that experimental design and analysis must be rigorous to be convincing, particularly when weak effects are claimed.