In field observations from a sinuous estuary, the drag coefficient (Formula presented.) based on the momentum balance was in the range of (Formula presented.), much greater than expected from bottom friction alone. (Formula presented.) also varied at tidal and seasonal timescales. (Formula presented.) was greater during flood tides than ebbs, most notably during spring tides. The ebb tide (Formula presented.) was negatively correlated with river discharge, while the flood tide (Formula presented.) showed no dependence on discharge. The large values of (Formula presented.) are explained by form drag from flow separation at sharp channel bends. Greater water depths during flood tides corresponded with increased values of (Formula presented.), consistent with the expected depth dependence for flow separation, as flow separation becomes stronger in deeper water. Additionally, the strength of the adverse pressure gradient downstream of the bend apex, which is indicative of flow separation, correlated with (Formula presented.) during flood tides. While (Formula presented.) generally increased with water depth, (Formula presented.) decreased for the highest water levels that corresponded with overbank flow. The decrease in (Formula presented.) may be due to the inhibition of flow separation with flow over the vegetated marsh. The dependence of (Formula presented.) during ebbs on discharge corresponds with the inhibition of flow separation by a favoring baroclinic pressure gradient that is locally generated at the bend apex due to curvature-induced secondary circulation. This effect increases with stratification, which increases with discharge. Additional factors may contribute to the high drag, including secondary circulation, multiple scales of bedforms, and shallow shoals, but the observations suggest that flow separation is the primary source.@en