Long-period teleseismic P waves recorded at VIC (Victoria, British Columbia) and COR (Corvallis, Orgeon) show anomalously large Ps conversions and later arriving P-to-S reverberations not observed from typical continental crustal sections or from previously proposed structures for these stations determined from refraction surveys. The timing and large amplitude of the Ps phase, relative to direct P, suggests a high velocity-contrast interface at 45 to 50 km depth under VIC and COR forming the base of a distinct low velocity zone. This interface is proposed to be the oceanic Moho which is being subducted under North America. Off azimuth Ps recorded at COR is consistent with a 20¿ eastward dip for the interface. Horizontal particle motion at both sites show evidence for lateral heterogeneity in local crustal structure. The distinct low velocity zone and its negative gradient with depth has important consequences for refraction interpretation in the region since the usual assumption of increasing velocity with depth is violated. Crustal thicknesses derived from such misinterpretation may be overestimated. In principle, this type of structure suggests a solution for the Vancouver Island crustal thickness problem in which the observed positive Bouguer gravity anomaly is inconsistent with the 50 km thick crustal thickness derived from previous refraction work.