Understanding, quantifying, and predicting wave transformation across marsh vegetation canopies is important for assessing the efficacy of nature-based shoreline strategies. A key challenge for modeling wave dissipation in coastal marshes is accurately representing the canopy drag, particularly as previously proposed canopy drag coefficient (CD) expressions vary considerably for the same vegetation and wave properties. There is also as yet no consensus on which dimensionless parameter best explains the variation of CD with wave conditions and water depth in marshes. This study addressed these challenges using wave, elevation, and vegetation measurements collected at two natural salt marshes across two summers. The dataset was used together with a wave energy conservation model to (i) identify and test possible reasons for differences among previously published CD expressions for salt marshes; and, (ii) derive and test expressions for CD as a function of different dimensionless parameters. We found that uncertainties in measurements of vegetation properties (shoot height, diameter, density) lead to substantial uncertainties in CD expressions, which may explain some differences among previous studies. When plotted versus Keulegan Carpenter number (KC) or Reynolds number (Re), CD values vary by roughly a factor of two for large waves (KC > 100) depending on water depth. When plotted versus Cauchy number (Ca), the ratio of hydrodynamic drag force to the canopy’s elastic restoring force, CD values collapse onto a single curve. We suggest that Ca is the more appropriate parameter for representing CD under larger waves because it represents the force balance relevant for plant motion and recommend future studies to test CD(Ca) relationships for different canopy heights.