Two methods which have been used in a number of studies of fluxes in salt marsh ecosystems, direct measurement by velocity-area estimation and hypsometrically-based volumetric simulation have been compared for calculating tidal flux in Tuff Crater. Tuff Crater, a mangrove-filled old volcanic explosion crater in the Waitemata Harbour, New Zealand, has a physiography particularly suitable for such a study, receiving minimal freshwater input, having clearly defined boundaries and with all tidal exchange occurring via the tidal creek. The topography of the basin was levelled, contouring at 5 cm intervals, with additional survey of creek cross-sections and a volume-stage curve was drawn. Discharge was estimated by changes in volume related to stage observations at the mouth of the basin. Direct measurement of flow was made at the gauging station by determining velocity at 0·2, 0·6 and 0·8 of the water depth and multiplying by cross-sectional area. Comparison of discharge-time curves using the two techniques reveals a similar pattern of discharge over time with peak ebb discharge exceeding peak flood discharge and peaks reaching similar values using the two techniques. However the volumetrically modelled discharge predicted an earlier reversal of the tide and an earlier ebb peak discharge. It is implied that the water surface is not horizontal at all tidal stages and that water surface slopes develop on the ebb tide resulting from inertial effects in restricted channels and on the mangrove-covered mudflats. The volumetric model is therefore inappropriate for the calculation of tidal flux, especially where this is used to calculate budgets of sediment or organic matter. The velocity-area method estimates discharges in excess of those predicted volumetrically. The disparity probably results from estimating discharge using only one station in the cross-section and is similar to errors of estimating discharge observed in other studies. © 1985.