We have studied 85 limestone and dolostone samples from 21 sites of the St. George and Table Head groups (Lower and Middle Ordovician) on Port au Port Peninsula (Humber Zone). Their gentle (~15°) tilt is probably Devonian. A steep viscous component dominant in the natural remanence is removed by thermal and (less efficiently) alternating-field demagnetization. In about 60% of all samples, the thermal treatment yielded a southeasterly reverse component ("A"), which typically remained blocked to 450–500 °C and then decayed to noise. Mean A component directions calculated from Zijderveld plots and referred to bedding are D, I = 148.5°, +33.6 °for N = 9 St. George sites and D, I = 150.0°, I = +40.7 °for N = 5 Table Head sites. Corresponding antipoles are 17.5°N, 152.3°E, dp, dm = 2.8°, 4.9 °and 13.4°N, 149.2°E, dp, dm = 2.9°, 4.7°, respectively. We interpret these results as representative of the geomagnetic field in the Early and mid-Ordovician. In 14 samples, mainly from the older St. George strata, thermal treatment yielded a reverse ("B") component with mean direction D, I = 167.9°, −9.2°, corresponding to an antipole at 44.9°N, 138.2°E, dp, dm = 3.6°, 7.1°. We interpret this as a late Paleozoic overprint in diagenetic hematite. The 30° anticlockwise rotation of Newfoundland proposed by Wegener is not supported by our data.