Multichannel seismic reflection data over the axial region of the East Pacific Rise are depth migrated using detailed velocity information from laboratory measurements of ophiolite samples. The data are from the Lamont Doherty Geological Observatory's line 17 shot across the rise at 9¿N. The migrated data are interpreted to show the structure of the seafloor and what we believe is the magma chamber roof. The polarity and apparent root mean squared velocity of the magma chamber roof are asymmetric with respect to the topographic axis of the ridge. The asymmetry is probably real and not an artifact of data collection. Modeling of the magnetic data gathered along the line shows that spreading has not been proceeding normally. A reasonable explanation for the asymmetry is the possibility that the line intersects a transform fault or an abandoned ridge segment near the ridge axis. The shape of the roof reflection is convex upward with an approximate slope of 10¿ and a width of 4 km. If extrapolated symmetrically to the other side of the ridge axis, the magma chamber roof event is consistent with the funnel-shaped chamber proposed by Pallister and Hopson (1981) for the Samail ophiolite. If the chamber roof steepens rapidly beyond the extent of the reflection, it would be consistent with the mush-filled model of Sleep (1975, 1978) and Dewey and Kidd (1977).