Two non-native clam species (Manila clams, Ruditapes philippinarum, and Eastern soft-shell clams, Mya arenaria) vary in density throughout intertidal habitats in Willapa Bay, Washington, USA. We used these species to test a fundamental ecological question, long addressed in marine systems, of whether the observed patterns were due to settlement choice or post-settlement mortality. A series of four experiments revealed habitat-specific mortality of young clams to be the dominant driver of distributions for both species. In contrast, settlement, which occurred over an extended reproductive period (June–Sept), appeared insensitive to substrate modification through additions of rock or shell (Expt 1–3). In Expt 2 and 3, sampled after winter when clams had reached ~1cm shell length, densities were lower in soft sediments characterizing much of the bay, and higher with rock or shell addition, especially with mesh bag protection. In Expt 4, which involved a crossed design of vegetation removal and substrate addition, one-year-old R. philippinarum densities increased with substrate addition more often than M. arenaria, and neither changed in response to removal of vegetation alone. In the case of one substratum, round rock, both vegetation removal and substrate addition were required to raise clam densities. Netting and gravel addition are standard practices in clam aquaculture to protect plantings of larger clams (e.g. >1cm), but in our study substrate modification reduced losses of smaller (<1mm) newly-settled clams. Overall, top-down control appears to be more important than settlement choice in determining local densities of Willapa Bay's non-native clams. Our findings open the possibility of improved aquaculture yields across all substrate types via natural settlement combined with interventions to reduce early losses due to predation.