On the northern Cascadia margin, single channel seismic data clearly image the bottom-simulating reflector (BSR) that marks the base of the hydrate stability zone. Airguns with three distinct source frequencies (30, 75, and 150 Hz) were recorded along several coincident lines, and at all frequencies the BSR appears as a single pulse. BSR reflection coefficients are slightly larger for the 30-Hz data than for the 75- or 150-Hz data. Multifrequency analysis indicates that the BSR is due to a single sharp interface at the base of the hydrate zone, whereas the velocity structure away from the interface must be gradationally-varying so as to produce no separate reflection and no tuning effects at the 75- and 150-Hz frequencies. In contrast, the seafloor reflection in some areas has significantly larger amplitudes for data recorded with the 75-Hz source, relative to the 30-Hz source. Synthetic seismogram analysis of the two data sets shows that the high amplitudes can be produced by a 2-m-thick high-velocity layer at the seafloor, with velocities and densities corresponding to those of carbonate. Using the 75-Hz source, seafloor and BSR amplitudes and reflection coefficients were measured over a tight grid of lines, nominally spaced at 200 m. Maximum BSR reflection coefficients of 0.15--0.18 were observed beneath topographic highs, while maximum seafloor reflection coefficients of up to 0.5--0.6 were found on the flanks of the topographic highs. This suggests that topography provides a major control on the flow of methane-bearing fluids. Amplitude measurements, converted to values of velocity above the BSR, were used to estimate hydrate concentrations, which reaches a maximum of 15--18% of the total sediment volume with mean values near 5--10%. Over the 100 km2 survey area, the volume of methane gas in the hydrate reservoir is estimated as 6.4¿1010 m3, or 2.1 trillion cubic feet. ¿ 1999 American Geophysical Union