Because secretive marsh birds are difficult to detect, population status and habitat use for these birds are not well known. We conducted repeated surveys for secretive marsh birds across 264 sites in the Upper Mississippi River and Great Lakes Joint Venture region to estimate abundance, occupancy, and detection probabilities during the 2008 and 2009 breeding seasons. We identified species groups based on observed species co-occurrences. Two species, least bittern Ixobrychus exilis and American bittern Botaurus lentiginosus, co-occurred with other species less often than expected by chance, and two species groups, rails (Virginia rail Rallus limicola and sora Porzana carolina) and open-water birds (pied-billed grebe Podilymbus podiceps, common moorhen Gallinula chloropus, and American coot Fulica americana; coots were only surveyed in 2009), co-occurred more often than expected by chance. These groupings were consistent between years. We then estimated the relation of these species and groups to landscape and local site characteristics by using zero-inflated abundance models that accounted for incomplete detection. At the landscape level (5-km radius), the amount of emergent herbaceous wetland was positively associated with least bittern occupancy, whereas the amount of woody wetland was negatively associated with least bittern, rail, and open-water bird occupancy. At the local level, habitat variables that were associated with abundance were not consistent among groups or between years, with the exception that both least bitterns and open-water birds had a strong positive association between abundance and water-vegetation interspersion. Land managers interested in marsh bird management or conservation may want to consider focusing efforts on landscapes with high amounts of emergent herbaceous wetland and low amounts of woody wetland, and managing for high amounts of water-vegetation interspersion within the wetland.