Hydraulic geometry and related analyses are often used to investigate tidal channel geometry and evolution and inform marsh restoration. An alternative approach is presented that avoids calculating tidal prism and allows analysis of additional channel metrics. It relies on scaling relationships between marsh island surface area and various metrics of the set of tidal channels draining each island. In the Skagit Delta marshes (Washington, United States), total channel surface area and length and surface area of the largest channel draining an island scaled disproportionately with island area, suggesting restoration of a 100-ha site would be preferable to restoration of 10 separate 10-ha sites to maximize channel length and area. A model of channel formation through random island conglomeration replicated observed scaling patterns, linking channel scaling to blind channel evolution from river distributaries. Channel size and complexity varied spatially, with significant deficits in an eroding marsh isolated from river distributaries and riverine sediments.