The salt marsh harvest mouse (SMHM; Reithrodontomys raviventris) is an endangered species restricted to the San Francisco Estuary wetlands. Approximately 90% of historical wetland habitat in the estuary has been destroyed. Modern wetlands primarily consist of relatively small, patchy fragments among a matrix of urban and upland habitat. Restoration and preservation of marsh habitat are among the primary methods of SMHM conservation, and understanding which patches of habitat to prioritize is key to allocating resources optimally. To this end, we developed a non-invasive survey for small mammals and used this technique to survey 48 marsh patches in the San Francisco Estuary. We tested hypotheses derived from island biogeography theory – that SMHM occupancy was related to marsh patch size, and inversely related to patch isolation (as measured by the proportion of marsh and urban habitat surrounding a given patch). We tested the effects of habitat variables and occurrence of potential competitors (western harvest mouse [R. megalotis], house mouse [Mus musculus], and California vole [Microtus californicus]) on SMHM occupancy. We also estimated patterns of occupancy in the three co-occurring species. Preliminary results suggest a strong relationship between SMHM occupancy and patch isolation at a landscape scale, and a moderate relationship with patch size. SMHM occupancy was also moderately increased by high tide escape habitat and inversely related to western harvest mouse occurrence. Western harvest mouse occupancy was positively related to the presence of terrestrial grasses within marsh patches. House mice and voles exhibited no strong habitat associations, although there was a moderate positive relationship between western harvest mouse and house mouse occupancy. Our results suggest that isolation, fragmentation, and encroachment of upland vegetation into marsh habitat are detrimental to SMHM persistence, and conservation of SMHM at the landscape scale may be particularly important.